diff --git "a/multi_student_7/checkpoint-9000/trainer_state.json" "b/multi_student_7/checkpoint-9000/trainer_state.json" new file mode 100644--- /dev/null +++ "b/multi_student_7/checkpoint-9000/trainer_state.json" @@ -0,0 +1,6334 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 4.632175032175033, + "eval_steps": 500, + "global_step": 9000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.005148005148005148, + "grad_norm": 19.51059658759793, + "learning_rate": 7.718696397941682e-07, + "loss": 8.2552, + "step": 10 + }, + { + "epoch": 0.010296010296010296, + "grad_norm": 19.70146681963939, + "learning_rate": 1.6295025728987994e-06, + "loss": 8.168, + "step": 20 + }, + { + "epoch": 0.015444015444015444, + "grad_norm": 9.795523599582376, + "learning_rate": 2.4871355060034306e-06, + "loss": 7.5953, + "step": 30 + }, + { + "epoch": 0.02059202059202059, + "grad_norm": 5.9480235723668144, + "learning_rate": 3.344768439108062e-06, + "loss": 6.8507, + "step": 40 + }, + { + "epoch": 0.02574002574002574, + "grad_norm": 3.5099463285447223, + "learning_rate": 4.202401372212693e-06, + "loss": 6.3398, + "step": 50 + }, + { + "epoch": 0.03088803088803089, + "grad_norm": 2.90890000793416, + "learning_rate": 5.060034305317325e-06, + "loss": 5.9988, + "step": 60 + }, + { + "epoch": 0.036036036036036036, + "grad_norm": 3.3978017812772445, + "learning_rate": 5.9176672384219555e-06, + "loss": 5.6503, + "step": 70 + }, + { + "epoch": 0.04118404118404118, + "grad_norm": 4.103088821630264, + "learning_rate": 6.775300171526587e-06, + "loss": 5.1289, + "step": 80 + }, + { + "epoch": 0.04633204633204633, + "grad_norm": 4.487842621805789, + "learning_rate": 7.632933104631218e-06, + "loss": 4.511, + "step": 90 + }, + { + "epoch": 0.05148005148005148, + "grad_norm": 7.491973587682773, + "learning_rate": 8.49056603773585e-06, + "loss": 3.9137, + "step": 100 + }, + { + "epoch": 0.05662805662805663, + "grad_norm": 7.323430680232261, + "learning_rate": 9.348198970840481e-06, + "loss": 3.4784, + "step": 110 + }, + { + "epoch": 0.06177606177606178, + "grad_norm": 6.619474879048035, + "learning_rate": 1.0205831903945113e-05, + "loss": 3.2067, + "step": 120 + }, + { + "epoch": 0.06692406692406692, + "grad_norm": 3.867444417417378, + "learning_rate": 1.1063464837049744e-05, + "loss": 3.0386, + "step": 130 + }, + { + "epoch": 0.07207207207207207, + "grad_norm": 4.117034943974073, + "learning_rate": 1.1921097770154374e-05, + "loss": 2.9159, + "step": 140 + }, + { + "epoch": 0.07722007722007722, + "grad_norm": 6.009222032591652, + "learning_rate": 1.2778730703259007e-05, + "loss": 2.796, + "step": 150 + }, + { + "epoch": 0.08236808236808237, + "grad_norm": 11.57505334622991, + "learning_rate": 1.3636363636363637e-05, + "loss": 2.7077, + "step": 160 + }, + { + "epoch": 0.08751608751608751, + "grad_norm": 11.428627421387615, + "learning_rate": 1.4493996569468268e-05, + "loss": 2.6403, + "step": 170 + }, + { + "epoch": 0.09266409266409266, + "grad_norm": 10.905546843716907, + "learning_rate": 1.53516295025729e-05, + "loss": 2.5735, + "step": 180 + }, + { + "epoch": 0.0978120978120978, + "grad_norm": 8.661551806496005, + "learning_rate": 1.620926243567753e-05, + "loss": 2.5078, + "step": 190 + }, + { + "epoch": 0.10296010296010295, + "grad_norm": 9.940151278870877, + "learning_rate": 1.7066895368782164e-05, + "loss": 2.4567, + "step": 200 + }, + { + "epoch": 0.10810810810810811, + "grad_norm": 13.091142904495205, + "learning_rate": 1.7924528301886792e-05, + "loss": 2.3953, + "step": 210 + }, + { + "epoch": 0.11325611325611326, + "grad_norm": 18.06611865183751, + "learning_rate": 1.8782161234991423e-05, + "loss": 2.329, + "step": 220 + }, + { + "epoch": 0.11840411840411841, + "grad_norm": 16.145152244084983, + "learning_rate": 1.9639794168096058e-05, + "loss": 2.2691, + "step": 230 + }, + { + "epoch": 0.12355212355212356, + "grad_norm": 35.1534299592601, + "learning_rate": 2.0497427101200686e-05, + "loss": 2.2524, + "step": 240 + }, + { + "epoch": 0.1287001287001287, + "grad_norm": 14.465207787142148, + "learning_rate": 2.1355060034305318e-05, + "loss": 2.1974, + "step": 250 + }, + { + "epoch": 0.13384813384813385, + "grad_norm": 11.353218963690644, + "learning_rate": 2.2212692967409952e-05, + "loss": 2.0873, + "step": 260 + }, + { + "epoch": 0.138996138996139, + "grad_norm": 12.277610068118626, + "learning_rate": 2.307032590051458e-05, + "loss": 2.0068, + "step": 270 + }, + { + "epoch": 0.14414414414414414, + "grad_norm": 10.505580540439455, + "learning_rate": 2.3927958833619212e-05, + "loss": 1.8888, + "step": 280 + }, + { + "epoch": 0.1492921492921493, + "grad_norm": 14.960676861450345, + "learning_rate": 2.4785591766723843e-05, + "loss": 1.7894, + "step": 290 + }, + { + "epoch": 0.15444015444015444, + "grad_norm": 8.613188849395, + "learning_rate": 2.564322469982847e-05, + "loss": 1.6967, + "step": 300 + }, + { + "epoch": 0.15958815958815958, + "grad_norm": 6.959841307105666, + "learning_rate": 2.6500857632933106e-05, + "loss": 1.6019, + "step": 310 + }, + { + "epoch": 0.16473616473616473, + "grad_norm": 5.864669793953841, + "learning_rate": 2.7358490566037738e-05, + "loss": 1.5326, + "step": 320 + }, + { + "epoch": 0.16988416988416988, + "grad_norm": 8.165848406130657, + "learning_rate": 2.8216123499142366e-05, + "loss": 1.4842, + "step": 330 + }, + { + "epoch": 0.17503217503217502, + "grad_norm": 6.747629987576382, + "learning_rate": 2.9073756432247e-05, + "loss": 1.426, + "step": 340 + }, + { + "epoch": 0.18018018018018017, + "grad_norm": 4.897733377675901, + "learning_rate": 2.9931389365351632e-05, + "loss": 1.3614, + "step": 350 + }, + { + "epoch": 0.18532818532818532, + "grad_norm": 5.330679969567318, + "learning_rate": 3.078902229845626e-05, + "loss": 1.3182, + "step": 360 + }, + { + "epoch": 0.19047619047619047, + "grad_norm": 4.722018507382775, + "learning_rate": 3.16466552315609e-05, + "loss": 1.26, + "step": 370 + }, + { + "epoch": 0.1956241956241956, + "grad_norm": 6.77470473058928, + "learning_rate": 3.250428816466552e-05, + "loss": 1.2128, + "step": 380 + }, + { + "epoch": 0.20077220077220076, + "grad_norm": 4.902136742839717, + "learning_rate": 3.3361921097770154e-05, + "loss": 1.1829, + "step": 390 + }, + { + "epoch": 0.2059202059202059, + "grad_norm": 5.515696505859524, + "learning_rate": 3.421955403087479e-05, + "loss": 1.1379, + "step": 400 + }, + { + "epoch": 0.21106821106821108, + "grad_norm": 3.9113735830318603, + "learning_rate": 3.507718696397942e-05, + "loss": 1.1002, + "step": 410 + }, + { + "epoch": 0.21621621621621623, + "grad_norm": 5.8061510583054545, + "learning_rate": 3.593481989708405e-05, + "loss": 1.0744, + "step": 420 + }, + { + "epoch": 0.22136422136422138, + "grad_norm": 5.927584495277828, + "learning_rate": 3.679245283018868e-05, + "loss": 1.0304, + "step": 430 + }, + { + "epoch": 0.22651222651222652, + "grad_norm": 5.920319878093127, + "learning_rate": 3.765008576329331e-05, + "loss": 1.0023, + "step": 440 + }, + { + "epoch": 0.23166023166023167, + "grad_norm": 5.302890318730487, + "learning_rate": 3.850771869639794e-05, + "loss": 0.9776, + "step": 450 + }, + { + "epoch": 0.23680823680823682, + "grad_norm": 5.653239199646973, + "learning_rate": 3.9365351629502574e-05, + "loss": 0.9482, + "step": 460 + }, + { + "epoch": 0.24195624195624196, + "grad_norm": 4.852345392975872, + "learning_rate": 4.0222984562607206e-05, + "loss": 0.9155, + "step": 470 + }, + { + "epoch": 0.2471042471042471, + "grad_norm": 4.499279652551338, + "learning_rate": 4.108061749571184e-05, + "loss": 0.8867, + "step": 480 + }, + { + "epoch": 0.25225225225225223, + "grad_norm": 5.181168954723113, + "learning_rate": 4.193825042881647e-05, + "loss": 0.8756, + "step": 490 + }, + { + "epoch": 0.2574002574002574, + "grad_norm": 6.255177791414999, + "learning_rate": 4.27958833619211e-05, + "loss": 0.8506, + "step": 500 + }, + { + "epoch": 0.2625482625482625, + "grad_norm": 5.396447906024421, + "learning_rate": 4.365351629502573e-05, + "loss": 0.8203, + "step": 510 + }, + { + "epoch": 0.2676962676962677, + "grad_norm": 7.371015885385536, + "learning_rate": 4.451114922813036e-05, + "loss": 0.8028, + "step": 520 + }, + { + "epoch": 0.2728442728442728, + "grad_norm": 6.198537777815005, + "learning_rate": 4.5368782161234994e-05, + "loss": 0.7808, + "step": 530 + }, + { + "epoch": 0.277992277992278, + "grad_norm": 4.485699286277638, + "learning_rate": 4.6226415094339625e-05, + "loss": 0.7639, + "step": 540 + }, + { + "epoch": 0.28314028314028317, + "grad_norm": 6.865323396700289, + "learning_rate": 4.708404802744426e-05, + "loss": 0.745, + "step": 550 + }, + { + "epoch": 0.2882882882882883, + "grad_norm": 4.718888462403314, + "learning_rate": 4.794168096054888e-05, + "loss": 0.7209, + "step": 560 + }, + { + "epoch": 0.29343629343629346, + "grad_norm": 4.887504164875748, + "learning_rate": 4.879931389365352e-05, + "loss": 0.7015, + "step": 570 + }, + { + "epoch": 0.2985842985842986, + "grad_norm": 5.372842102076737, + "learning_rate": 4.965694682675815e-05, + "loss": 0.6767, + "step": 580 + }, + { + "epoch": 0.30373230373230375, + "grad_norm": 6.766330192939623, + "learning_rate": 5.0514579759862776e-05, + "loss": 0.6624, + "step": 590 + }, + { + "epoch": 0.3088803088803089, + "grad_norm": 4.614837742067113, + "learning_rate": 5.1372212692967414e-05, + "loss": 0.6405, + "step": 600 + }, + { + "epoch": 0.31402831402831405, + "grad_norm": 4.201337449433587, + "learning_rate": 5.222984562607204e-05, + "loss": 0.6194, + "step": 610 + }, + { + "epoch": 0.31917631917631917, + "grad_norm": 5.51280432030841, + "learning_rate": 5.308747855917667e-05, + "loss": 0.5951, + "step": 620 + }, + { + "epoch": 0.32432432432432434, + "grad_norm": 5.129963269925849, + "learning_rate": 5.394511149228131e-05, + "loss": 0.5777, + "step": 630 + }, + { + "epoch": 0.32947232947232946, + "grad_norm": 4.412209442955385, + "learning_rate": 5.480274442538593e-05, + "loss": 0.5642, + "step": 640 + }, + { + "epoch": 0.33462033462033464, + "grad_norm": 4.336638446576286, + "learning_rate": 5.5660377358490564e-05, + "loss": 0.5426, + "step": 650 + }, + { + "epoch": 0.33976833976833976, + "grad_norm": 3.556283909809826, + "learning_rate": 5.65180102915952e-05, + "loss": 0.5112, + "step": 660 + }, + { + "epoch": 0.34491634491634493, + "grad_norm": 5.225584058992361, + "learning_rate": 5.737564322469983e-05, + "loss": 0.5044, + "step": 670 + }, + { + "epoch": 0.35006435006435005, + "grad_norm": 3.1697056899798186, + "learning_rate": 5.823327615780446e-05, + "loss": 0.4857, + "step": 680 + }, + { + "epoch": 0.3552123552123552, + "grad_norm": 3.5258824608431003, + "learning_rate": 5.90909090909091e-05, + "loss": 0.4747, + "step": 690 + }, + { + "epoch": 0.36036036036036034, + "grad_norm": 3.5240284197467155, + "learning_rate": 5.994854202401372e-05, + "loss": 0.4563, + "step": 700 + }, + { + "epoch": 0.3655083655083655, + "grad_norm": 3.787588207851851, + "learning_rate": 6.080617495711835e-05, + "loss": 0.4382, + "step": 710 + }, + { + "epoch": 0.37065637065637064, + "grad_norm": 3.28493026844208, + "learning_rate": 6.166380789022299e-05, + "loss": 0.4216, + "step": 720 + }, + { + "epoch": 0.3758043758043758, + "grad_norm": 3.144221026231138, + "learning_rate": 6.252144082332761e-05, + "loss": 0.417, + "step": 730 + }, + { + "epoch": 0.38095238095238093, + "grad_norm": 2.780102841682132, + "learning_rate": 6.337907375643225e-05, + "loss": 0.3942, + "step": 740 + }, + { + "epoch": 0.3861003861003861, + "grad_norm": 3.1996820882351713, + "learning_rate": 6.423670668953689e-05, + "loss": 0.3787, + "step": 750 + }, + { + "epoch": 0.3912483912483912, + "grad_norm": 2.8374055946677923, + "learning_rate": 6.50943396226415e-05, + "loss": 0.3714, + "step": 760 + }, + { + "epoch": 0.3963963963963964, + "grad_norm": 3.0094036046611454, + "learning_rate": 6.595197255574615e-05, + "loss": 0.3606, + "step": 770 + }, + { + "epoch": 0.4015444015444015, + "grad_norm": 2.6679117500690204, + "learning_rate": 6.680960548885078e-05, + "loss": 0.3379, + "step": 780 + }, + { + "epoch": 0.4066924066924067, + "grad_norm": 3.101181469988925, + "learning_rate": 6.76672384219554e-05, + "loss": 0.3271, + "step": 790 + }, + { + "epoch": 0.4118404118404118, + "grad_norm": 2.421653916410696, + "learning_rate": 6.852487135506004e-05, + "loss": 0.3132, + "step": 800 + }, + { + "epoch": 0.416988416988417, + "grad_norm": 2.1110958003741174, + "learning_rate": 6.938250428816467e-05, + "loss": 0.3013, + "step": 810 + }, + { + "epoch": 0.42213642213642216, + "grad_norm": 1.9804288224535966, + "learning_rate": 7.024013722126929e-05, + "loss": 0.2892, + "step": 820 + }, + { + "epoch": 0.4272844272844273, + "grad_norm": 2.0965122713743725, + "learning_rate": 7.109777015437394e-05, + "loss": 0.2802, + "step": 830 + }, + { + "epoch": 0.43243243243243246, + "grad_norm": 2.0662383619585323, + "learning_rate": 7.195540308747857e-05, + "loss": 0.2666, + "step": 840 + }, + { + "epoch": 0.4375804375804376, + "grad_norm": 1.9080845115386118, + "learning_rate": 7.281303602058319e-05, + "loss": 0.2557, + "step": 850 + }, + { + "epoch": 0.44272844272844275, + "grad_norm": 2.435492621224063, + "learning_rate": 7.367066895368783e-05, + "loss": 0.2441, + "step": 860 + }, + { + "epoch": 0.44787644787644787, + "grad_norm": 2.3478908862806707, + "learning_rate": 7.452830188679245e-05, + "loss": 0.2302, + "step": 870 + }, + { + "epoch": 0.45302445302445304, + "grad_norm": 2.033330789035785, + "learning_rate": 7.538593481989708e-05, + "loss": 0.222, + "step": 880 + }, + { + "epoch": 0.45817245817245816, + "grad_norm": 1.8373992771294045, + "learning_rate": 7.624356775300173e-05, + "loss": 0.2079, + "step": 890 + }, + { + "epoch": 0.46332046332046334, + "grad_norm": 1.9518420548134319, + "learning_rate": 7.710120068610634e-05, + "loss": 0.1979, + "step": 900 + }, + { + "epoch": 0.46846846846846846, + "grad_norm": 1.6103496909038504, + "learning_rate": 7.795883361921097e-05, + "loss": 0.1936, + "step": 910 + }, + { + "epoch": 0.47361647361647363, + "grad_norm": 1.3694934489721085, + "learning_rate": 7.881646655231562e-05, + "loss": 0.1841, + "step": 920 + }, + { + "epoch": 0.47876447876447875, + "grad_norm": 1.2736699951465489, + "learning_rate": 7.967409948542024e-05, + "loss": 0.1766, + "step": 930 + }, + { + "epoch": 0.4839124839124839, + "grad_norm": 2.1829922501226444, + "learning_rate": 8.053173241852487e-05, + "loss": 0.1675, + "step": 940 + }, + { + "epoch": 0.48906048906048905, + "grad_norm": 1.659964942426995, + "learning_rate": 8.138936535162951e-05, + "loss": 0.161, + "step": 950 + }, + { + "epoch": 0.4942084942084942, + "grad_norm": 1.5767630911351294, + "learning_rate": 8.224699828473413e-05, + "loss": 0.1494, + "step": 960 + }, + { + "epoch": 0.49935649935649934, + "grad_norm": 1.537782758985047, + "learning_rate": 8.310463121783876e-05, + "loss": 0.1445, + "step": 970 + }, + { + "epoch": 0.5045045045045045, + "grad_norm": 1.604101659458472, + "learning_rate": 8.396226415094341e-05, + "loss": 0.1402, + "step": 980 + }, + { + "epoch": 0.5096525096525096, + "grad_norm": 2.256805925155236, + "learning_rate": 8.481989708404803e-05, + "loss": 0.1405, + "step": 990 + }, + { + "epoch": 0.5148005148005148, + "grad_norm": 1.4857578289963005, + "learning_rate": 8.567753001715266e-05, + "loss": 0.1321, + "step": 1000 + }, + { + "epoch": 0.51994851994852, + "grad_norm": 1.0326983623178083, + "learning_rate": 8.65351629502573e-05, + "loss": 0.1224, + "step": 1010 + }, + { + "epoch": 0.525096525096525, + "grad_norm": 1.1180976317039228, + "learning_rate": 8.739279588336192e-05, + "loss": 0.1189, + "step": 1020 + }, + { + "epoch": 0.5302445302445302, + "grad_norm": 1.1721351234336, + "learning_rate": 8.825042881646655e-05, + "loss": 0.1092, + "step": 1030 + }, + { + "epoch": 0.5353925353925354, + "grad_norm": 0.9640093738745075, + "learning_rate": 8.91080617495712e-05, + "loss": 0.1021, + "step": 1040 + }, + { + "epoch": 0.5405405405405406, + "grad_norm": 1.0306043643976568, + "learning_rate": 8.996569468267581e-05, + "loss": 0.0982, + "step": 1050 + }, + { + "epoch": 0.5456885456885456, + "grad_norm": 1.4999347000799717, + "learning_rate": 9.082332761578045e-05, + "loss": 0.0936, + "step": 1060 + }, + { + "epoch": 0.5508365508365508, + "grad_norm": 1.0452281614309868, + "learning_rate": 9.168096054888509e-05, + "loss": 0.0878, + "step": 1070 + }, + { + "epoch": 0.555984555984556, + "grad_norm": 1.2088207649167284, + "learning_rate": 9.253859348198971e-05, + "loss": 0.0824, + "step": 1080 + }, + { + "epoch": 0.5611325611325612, + "grad_norm": 0.6373898488496081, + "learning_rate": 9.339622641509434e-05, + "loss": 0.0809, + "step": 1090 + }, + { + "epoch": 0.5662805662805663, + "grad_norm": 0.5917955695640877, + "learning_rate": 9.425385934819899e-05, + "loss": 0.0742, + "step": 1100 + }, + { + "epoch": 0.5714285714285714, + "grad_norm": 0.6146276278340692, + "learning_rate": 9.51114922813036e-05, + "loss": 0.0706, + "step": 1110 + }, + { + "epoch": 0.5765765765765766, + "grad_norm": 0.8766352471472525, + "learning_rate": 9.596912521440823e-05, + "loss": 0.0672, + "step": 1120 + }, + { + "epoch": 0.5817245817245817, + "grad_norm": 0.6678260178109867, + "learning_rate": 9.682675814751287e-05, + "loss": 0.0655, + "step": 1130 + }, + { + "epoch": 0.5868725868725869, + "grad_norm": 0.49817357185278055, + "learning_rate": 9.76843910806175e-05, + "loss": 0.0611, + "step": 1140 + }, + { + "epoch": 0.592020592020592, + "grad_norm": 1.0087468424307855, + "learning_rate": 9.854202401372213e-05, + "loss": 0.0611, + "step": 1150 + }, + { + "epoch": 0.5971685971685972, + "grad_norm": 0.5362523083451194, + "learning_rate": 9.939965694682676e-05, + "loss": 0.0579, + "step": 1160 + }, + { + "epoch": 0.6023166023166023, + "grad_norm": 0.4534299709647709, + "learning_rate": 9.99999798272238e-05, + "loss": 0.0549, + "step": 1170 + }, + { + "epoch": 0.6074646074646075, + "grad_norm": 0.6406776231632558, + "learning_rate": 9.999962120054416e-05, + "loss": 0.0534, + "step": 1180 + }, + { + "epoch": 0.6126126126126126, + "grad_norm": 0.5792838241902996, + "learning_rate": 9.999881429364988e-05, + "loss": 0.0512, + "step": 1190 + }, + { + "epoch": 0.6177606177606177, + "grad_norm": 0.4275795011357426, + "learning_rate": 9.999755911377544e-05, + "loss": 0.0488, + "step": 1200 + }, + { + "epoch": 0.6229086229086229, + "grad_norm": 0.33615545845119615, + "learning_rate": 9.999585567217439e-05, + "loss": 0.0476, + "step": 1210 + }, + { + "epoch": 0.6280566280566281, + "grad_norm": 0.33655008372027123, + "learning_rate": 9.99937039841192e-05, + "loss": 0.0444, + "step": 1220 + }, + { + "epoch": 0.6332046332046332, + "grad_norm": 0.43244236090431704, + "learning_rate": 9.999110406890121e-05, + "loss": 0.0438, + "step": 1230 + }, + { + "epoch": 0.6383526383526383, + "grad_norm": 0.41139634543075815, + "learning_rate": 9.998805594983043e-05, + "loss": 0.042, + "step": 1240 + }, + { + "epoch": 0.6435006435006435, + "grad_norm": 0.5206972896361842, + "learning_rate": 9.998455965423525e-05, + "loss": 0.0431, + "step": 1250 + }, + { + "epoch": 0.6486486486486487, + "grad_norm": 0.3937459395109014, + "learning_rate": 9.998061521346236e-05, + "loss": 0.041, + "step": 1260 + }, + { + "epoch": 0.6537966537966537, + "grad_norm": 0.2963678010736684, + "learning_rate": 9.997622266287629e-05, + "loss": 0.0393, + "step": 1270 + }, + { + "epoch": 0.6589446589446589, + "grad_norm": 0.26754815540836463, + "learning_rate": 9.997138204185923e-05, + "loss": 0.0376, + "step": 1280 + }, + { + "epoch": 0.6640926640926641, + "grad_norm": 0.29045330821553156, + "learning_rate": 9.996609339381057e-05, + "loss": 0.036, + "step": 1290 + }, + { + "epoch": 0.6692406692406693, + "grad_norm": 0.3089797983346258, + "learning_rate": 9.996035676614663e-05, + "loss": 0.0356, + "step": 1300 + }, + { + "epoch": 0.6743886743886743, + "grad_norm": 0.26639822964562176, + "learning_rate": 9.99541722103001e-05, + "loss": 0.0349, + "step": 1310 + }, + { + "epoch": 0.6795366795366795, + "grad_norm": 0.45121223083554485, + "learning_rate": 9.994753978171971e-05, + "loss": 0.0351, + "step": 1320 + }, + { + "epoch": 0.6846846846846847, + "grad_norm": 0.3667468018122777, + "learning_rate": 9.994045953986964e-05, + "loss": 0.0338, + "step": 1330 + }, + { + "epoch": 0.6898326898326899, + "grad_norm": 0.24841154017565364, + "learning_rate": 9.993293154822901e-05, + "loss": 0.0328, + "step": 1340 + }, + { + "epoch": 0.694980694980695, + "grad_norm": 0.283360706082952, + "learning_rate": 9.992495587429129e-05, + "loss": 0.0315, + "step": 1350 + }, + { + "epoch": 0.7001287001287001, + "grad_norm": 0.24413900296621255, + "learning_rate": 9.991653258956381e-05, + "loss": 0.0311, + "step": 1360 + }, + { + "epoch": 0.7052767052767053, + "grad_norm": 0.28975076041601167, + "learning_rate": 9.990766176956697e-05, + "loss": 0.0303, + "step": 1370 + }, + { + "epoch": 0.7104247104247104, + "grad_norm": 0.26351763932473593, + "learning_rate": 9.989834349383361e-05, + "loss": 0.03, + "step": 1380 + }, + { + "epoch": 0.7155727155727156, + "grad_norm": 0.5331124737621773, + "learning_rate": 9.988857784590836e-05, + "loss": 0.0291, + "step": 1390 + }, + { + "epoch": 0.7207207207207207, + "grad_norm": 0.24186391898474244, + "learning_rate": 9.98783649133468e-05, + "loss": 0.0284, + "step": 1400 + }, + { + "epoch": 0.7258687258687259, + "grad_norm": 0.21615992134341075, + "learning_rate": 9.986770478771475e-05, + "loss": 0.0273, + "step": 1410 + }, + { + "epoch": 0.731016731016731, + "grad_norm": 0.21264572442174073, + "learning_rate": 9.985659756458741e-05, + "loss": 0.0267, + "step": 1420 + }, + { + "epoch": 0.7361647361647362, + "grad_norm": 0.2320920564652625, + "learning_rate": 9.984504334354847e-05, + "loss": 0.0262, + "step": 1430 + }, + { + "epoch": 0.7413127413127413, + "grad_norm": 0.2342099433653173, + "learning_rate": 9.983304222818929e-05, + "loss": 0.0252, + "step": 1440 + }, + { + "epoch": 0.7464607464607464, + "grad_norm": 0.240834162772985, + "learning_rate": 9.9820594326108e-05, + "loss": 0.0245, + "step": 1450 + }, + { + "epoch": 0.7516087516087516, + "grad_norm": 0.2161277200650416, + "learning_rate": 9.980769974890833e-05, + "loss": 0.0238, + "step": 1460 + }, + { + "epoch": 0.7567567567567568, + "grad_norm": 0.24019983928841215, + "learning_rate": 9.979435861219888e-05, + "loss": 0.0233, + "step": 1470 + }, + { + "epoch": 0.7619047619047619, + "grad_norm": 0.2970865898808262, + "learning_rate": 9.978057103559188e-05, + "loss": 0.0226, + "step": 1480 + }, + { + "epoch": 0.767052767052767, + "grad_norm": 0.24780480301331292, + "learning_rate": 9.976633714270226e-05, + "loss": 0.0221, + "step": 1490 + }, + { + "epoch": 0.7722007722007722, + "grad_norm": 0.25727613523882426, + "learning_rate": 9.975165706114639e-05, + "loss": 0.0214, + "step": 1500 + }, + { + "epoch": 0.7773487773487774, + "grad_norm": 0.24359971967254024, + "learning_rate": 9.973653092254113e-05, + "loss": 0.0209, + "step": 1510 + }, + { + "epoch": 0.7824967824967825, + "grad_norm": 0.2067746721639629, + "learning_rate": 9.97209588625024e-05, + "loss": 0.0202, + "step": 1520 + }, + { + "epoch": 0.7876447876447876, + "grad_norm": 0.19304855106218685, + "learning_rate": 9.970494102064427e-05, + "loss": 0.0195, + "step": 1530 + }, + { + "epoch": 0.7927927927927928, + "grad_norm": 0.22693160915077307, + "learning_rate": 9.968847754057742e-05, + "loss": 0.019, + "step": 1540 + }, + { + "epoch": 0.797940797940798, + "grad_norm": 0.23060204821027103, + "learning_rate": 9.967156856990801e-05, + "loss": 0.0186, + "step": 1550 + }, + { + "epoch": 0.803088803088803, + "grad_norm": 2.572239383211685, + "learning_rate": 9.965421426023631e-05, + "loss": 0.1139, + "step": 1560 + }, + { + "epoch": 0.8082368082368082, + "grad_norm": 0.5437887934638203, + "learning_rate": 9.963641476715541e-05, + "loss": 0.0343, + "step": 1570 + }, + { + "epoch": 0.8133848133848134, + "grad_norm": 0.23796250360876128, + "learning_rate": 9.961817025024974e-05, + "loss": 0.0235, + "step": 1580 + }, + { + "epoch": 0.8185328185328186, + "grad_norm": 0.1764844020309923, + "learning_rate": 9.959948087309362e-05, + "loss": 0.0195, + "step": 1590 + }, + { + "epoch": 0.8236808236808236, + "grad_norm": 0.17189189871306304, + "learning_rate": 9.95803468032499e-05, + "loss": 0.0177, + "step": 1600 + }, + { + "epoch": 0.8288288288288288, + "grad_norm": 0.17751312710898323, + "learning_rate": 9.956076821226836e-05, + "loss": 0.0171, + "step": 1610 + }, + { + "epoch": 0.833976833976834, + "grad_norm": 0.16437143945406296, + "learning_rate": 9.954074527568426e-05, + "loss": 0.0161, + "step": 1620 + }, + { + "epoch": 0.8391248391248392, + "grad_norm": 0.15832400304822497, + "learning_rate": 9.952027817301667e-05, + "loss": 0.0157, + "step": 1630 + }, + { + "epoch": 0.8442728442728443, + "grad_norm": 0.16117748572605023, + "learning_rate": 9.949936708776691e-05, + "loss": 0.0148, + "step": 1640 + }, + { + "epoch": 0.8494208494208494, + "grad_norm": 0.16984391534714557, + "learning_rate": 9.947801220741695e-05, + "loss": 0.0143, + "step": 1650 + }, + { + "epoch": 0.8545688545688546, + "grad_norm": 0.1696223065684197, + "learning_rate": 9.945621372342762e-05, + "loss": 0.0137, + "step": 1660 + }, + { + "epoch": 0.8597168597168597, + "grad_norm": 0.1800035242900872, + "learning_rate": 9.943397183123697e-05, + "loss": 0.0133, + "step": 1670 + }, + { + "epoch": 0.8648648648648649, + "grad_norm": 0.15249239519006993, + "learning_rate": 9.941128673025854e-05, + "loss": 0.0133, + "step": 1680 + }, + { + "epoch": 0.87001287001287, + "grad_norm": 0.1551659031233157, + "learning_rate": 9.938815862387953e-05, + "loss": 0.0132, + "step": 1690 + }, + { + "epoch": 0.8751608751608752, + "grad_norm": 0.14348975402940561, + "learning_rate": 9.936458771945891e-05, + "loss": 0.0131, + "step": 1700 + }, + { + "epoch": 0.8803088803088803, + "grad_norm": 0.140601933336413, + "learning_rate": 9.934057422832572e-05, + "loss": 0.0123, + "step": 1710 + }, + { + "epoch": 0.8854568854568855, + "grad_norm": 0.1275666846983677, + "learning_rate": 9.931611836577703e-05, + "loss": 0.012, + "step": 1720 + }, + { + "epoch": 0.8906048906048906, + "grad_norm": 0.13327069275308084, + "learning_rate": 9.92912203510761e-05, + "loss": 0.012, + "step": 1730 + }, + { + "epoch": 0.8957528957528957, + "grad_norm": 0.1658754837807317, + "learning_rate": 9.926588040745035e-05, + "loss": 0.0117, + "step": 1740 + }, + { + "epoch": 0.9009009009009009, + "grad_norm": 0.14050126640493185, + "learning_rate": 9.924009876208944e-05, + "loss": 0.0116, + "step": 1750 + }, + { + "epoch": 0.9060489060489061, + "grad_norm": 0.1413904681709387, + "learning_rate": 9.921387564614313e-05, + "loss": 0.0111, + "step": 1760 + }, + { + "epoch": 0.9111969111969112, + "grad_norm": 0.13671787855200412, + "learning_rate": 9.918721129471927e-05, + "loss": 0.0113, + "step": 1770 + }, + { + "epoch": 0.9163449163449163, + "grad_norm": 0.1300287060008007, + "learning_rate": 9.916010594688173e-05, + "loss": 0.011, + "step": 1780 + }, + { + "epoch": 0.9214929214929215, + "grad_norm": 0.12272142768731135, + "learning_rate": 9.913255984564814e-05, + "loss": 0.0104, + "step": 1790 + }, + { + "epoch": 0.9266409266409267, + "grad_norm": 0.13585333911080333, + "learning_rate": 9.910457323798783e-05, + "loss": 0.0101, + "step": 1800 + }, + { + "epoch": 0.9317889317889317, + "grad_norm": 0.1365832152868149, + "learning_rate": 9.907614637481957e-05, + "loss": 0.0104, + "step": 1810 + }, + { + "epoch": 0.9369369369369369, + "grad_norm": 0.13886380017866182, + "learning_rate": 9.904727951100926e-05, + "loss": 0.0099, + "step": 1820 + }, + { + "epoch": 0.9420849420849421, + "grad_norm": 0.13559410476056236, + "learning_rate": 9.901797290536775e-05, + "loss": 0.0098, + "step": 1830 + }, + { + "epoch": 0.9472329472329473, + "grad_norm": 0.1380086457171486, + "learning_rate": 9.898822682064848e-05, + "loss": 0.0098, + "step": 1840 + }, + { + "epoch": 0.9523809523809523, + "grad_norm": 0.6177427987927422, + "learning_rate": 9.895804152354507e-05, + "loss": 0.0098, + "step": 1850 + }, + { + "epoch": 0.9575289575289575, + "grad_norm": 0.508965354576064, + "learning_rate": 9.892741728468902e-05, + "loss": 0.0154, + "step": 1860 + }, + { + "epoch": 0.9626769626769627, + "grad_norm": 0.1734010623999643, + "learning_rate": 9.889635437864719e-05, + "loss": 0.0125, + "step": 1870 + }, + { + "epoch": 0.9678249678249679, + "grad_norm": 0.15703615270146173, + "learning_rate": 9.886485308391944e-05, + "loss": 0.0104, + "step": 1880 + }, + { + "epoch": 0.972972972972973, + "grad_norm": 0.14238560159782349, + "learning_rate": 9.883291368293605e-05, + "loss": 0.0093, + "step": 1890 + }, + { + "epoch": 0.9781209781209781, + "grad_norm": 0.11593480562868673, + "learning_rate": 9.880053646205523e-05, + "loss": 0.0088, + "step": 1900 + }, + { + "epoch": 0.9832689832689833, + "grad_norm": 0.1250311748465019, + "learning_rate": 9.87677217115605e-05, + "loss": 0.0087, + "step": 1910 + }, + { + "epoch": 0.9884169884169884, + "grad_norm": 0.10915721832834366, + "learning_rate": 9.873446972565816e-05, + "loss": 0.0084, + "step": 1920 + }, + { + "epoch": 0.9935649935649936, + "grad_norm": 0.11826713704360893, + "learning_rate": 9.870078080247464e-05, + "loss": 0.0084, + "step": 1930 + }, + { + "epoch": 0.9987129987129987, + "grad_norm": 0.13132056186197924, + "learning_rate": 9.866665524405377e-05, + "loss": 0.0082, + "step": 1940 + }, + { + "epoch": 1.0036036036036036, + "grad_norm": 0.11858168124629603, + "learning_rate": 9.86320933563541e-05, + "loss": 0.0077, + "step": 1950 + }, + { + "epoch": 1.0087516087516089, + "grad_norm": 0.1237462850171375, + "learning_rate": 9.859709544924624e-05, + "loss": 0.0076, + "step": 1960 + }, + { + "epoch": 1.013899613899614, + "grad_norm": 0.1224220645410722, + "learning_rate": 9.85616618365099e-05, + "loss": 0.0075, + "step": 1970 + }, + { + "epoch": 1.019047619047619, + "grad_norm": 0.13079765571571764, + "learning_rate": 9.852579283583126e-05, + "loss": 0.0075, + "step": 1980 + }, + { + "epoch": 1.0241956241956243, + "grad_norm": 0.11011593961405876, + "learning_rate": 9.848948876880002e-05, + "loss": 0.0073, + "step": 1990 + }, + { + "epoch": 1.0293436293436293, + "grad_norm": 0.10429883248159245, + "learning_rate": 9.845274996090655e-05, + "loss": 0.007, + "step": 2000 + }, + { + "epoch": 1.0344916344916344, + "grad_norm": 0.10883414844159683, + "learning_rate": 9.841557674153897e-05, + "loss": 0.0069, + "step": 2010 + }, + { + "epoch": 1.0396396396396397, + "grad_norm": 0.10985446797058561, + "learning_rate": 9.837796944398014e-05, + "loss": 0.0068, + "step": 2020 + }, + { + "epoch": 1.0447876447876447, + "grad_norm": 0.10929404720847335, + "learning_rate": 9.833992840540481e-05, + "loss": 0.0069, + "step": 2030 + }, + { + "epoch": 1.04993564993565, + "grad_norm": 0.10611541757073792, + "learning_rate": 9.830145396687641e-05, + "loss": 0.0068, + "step": 2040 + }, + { + "epoch": 1.055083655083655, + "grad_norm": 0.09681663896670493, + "learning_rate": 9.826254647334419e-05, + "loss": 0.0066, + "step": 2050 + }, + { + "epoch": 1.0602316602316602, + "grad_norm": 0.10666808939263758, + "learning_rate": 9.822320627363997e-05, + "loss": 0.0064, + "step": 2060 + }, + { + "epoch": 1.0653796653796654, + "grad_norm": 0.11874311376321617, + "learning_rate": 9.81834337204751e-05, + "loss": 0.0065, + "step": 2070 + }, + { + "epoch": 1.0705276705276705, + "grad_norm": 0.10330219941649453, + "learning_rate": 9.814322917043721e-05, + "loss": 0.0065, + "step": 2080 + }, + { + "epoch": 1.0756756756756758, + "grad_norm": 0.10518179079688333, + "learning_rate": 9.810259298398717e-05, + "loss": 0.0063, + "step": 2090 + }, + { + "epoch": 1.0808236808236809, + "grad_norm": 0.11196650081811249, + "learning_rate": 9.806152552545567e-05, + "loss": 0.0061, + "step": 2100 + }, + { + "epoch": 1.085971685971686, + "grad_norm": 0.10391144612302242, + "learning_rate": 9.80200271630401e-05, + "loss": 0.0063, + "step": 2110 + }, + { + "epoch": 1.0911196911196912, + "grad_norm": 0.10437009383950677, + "learning_rate": 9.797809826880118e-05, + "loss": 0.006, + "step": 2120 + }, + { + "epoch": 1.0962676962676963, + "grad_norm": 0.15742816962279707, + "learning_rate": 9.793573921865961e-05, + "loss": 0.0061, + "step": 2130 + }, + { + "epoch": 1.1014157014157013, + "grad_norm": 1.2391996261097045, + "learning_rate": 9.789295039239277e-05, + "loss": 0.009, + "step": 2140 + }, + { + "epoch": 1.1065637065637066, + "grad_norm": 0.4258283630396214, + "learning_rate": 9.784973217363123e-05, + "loss": 0.0167, + "step": 2150 + }, + { + "epoch": 1.1117117117117117, + "grad_norm": 0.22704282205376575, + "learning_rate": 9.78060849498554e-05, + "loss": 0.011, + "step": 2160 + }, + { + "epoch": 1.1168597168597167, + "grad_norm": 0.2903222759151862, + "learning_rate": 9.776200911239196e-05, + "loss": 0.0146, + "step": 2170 + }, + { + "epoch": 1.122007722007722, + "grad_norm": 0.13920400223191687, + "learning_rate": 9.771750505641044e-05, + "loss": 0.0091, + "step": 2180 + }, + { + "epoch": 1.127155727155727, + "grad_norm": 0.11920796398066835, + "learning_rate": 9.767257318091961e-05, + "loss": 0.0074, + "step": 2190 + }, + { + "epoch": 1.1323037323037324, + "grad_norm": 0.10527324354349751, + "learning_rate": 9.76272138887639e-05, + "loss": 0.0068, + "step": 2200 + }, + { + "epoch": 1.1374517374517374, + "grad_norm": 0.09156854644361, + "learning_rate": 9.758142758661992e-05, + "loss": 0.0062, + "step": 2210 + }, + { + "epoch": 1.1425997425997425, + "grad_norm": 0.10256933439724165, + "learning_rate": 9.753521468499261e-05, + "loss": 0.0058, + "step": 2220 + }, + { + "epoch": 1.1477477477477478, + "grad_norm": 0.09728084601303465, + "learning_rate": 9.748857559821171e-05, + "loss": 0.0059, + "step": 2230 + }, + { + "epoch": 1.1528957528957529, + "grad_norm": 0.09070888908222362, + "learning_rate": 9.744151074442798e-05, + "loss": 0.0056, + "step": 2240 + }, + { + "epoch": 1.1580437580437581, + "grad_norm": 0.08815351818450588, + "learning_rate": 9.739402054560948e-05, + "loss": 0.0054, + "step": 2250 + }, + { + "epoch": 1.1631917631917632, + "grad_norm": 0.08362202696869882, + "learning_rate": 9.734610542753775e-05, + "loss": 0.0052, + "step": 2260 + }, + { + "epoch": 1.1683397683397683, + "grad_norm": 0.08985409051259748, + "learning_rate": 9.729776581980403e-05, + "loss": 0.0054, + "step": 2270 + }, + { + "epoch": 1.1734877734877736, + "grad_norm": 0.09519354418158256, + "learning_rate": 9.724900215580542e-05, + "loss": 0.0052, + "step": 2280 + }, + { + "epoch": 1.1786357786357786, + "grad_norm": 0.09918504442751083, + "learning_rate": 9.719981487274094e-05, + "loss": 0.0054, + "step": 2290 + }, + { + "epoch": 1.1837837837837837, + "grad_norm": 0.08289788258009483, + "learning_rate": 9.715020441160764e-05, + "loss": 0.0053, + "step": 2300 + }, + { + "epoch": 1.188931788931789, + "grad_norm": 0.09232435159305605, + "learning_rate": 9.710017121719666e-05, + "loss": 0.0049, + "step": 2310 + }, + { + "epoch": 1.194079794079794, + "grad_norm": 0.08339859786132367, + "learning_rate": 9.704971573808921e-05, + "loss": 0.0049, + "step": 2320 + }, + { + "epoch": 1.1992277992277993, + "grad_norm": 0.08650033957989789, + "learning_rate": 9.699883842665258e-05, + "loss": 0.0048, + "step": 2330 + }, + { + "epoch": 1.2043758043758044, + "grad_norm": 0.0794400511956092, + "learning_rate": 9.694753973903608e-05, + "loss": 0.0049, + "step": 2340 + }, + { + "epoch": 1.2095238095238094, + "grad_norm": 0.0902066930501087, + "learning_rate": 9.689582013516692e-05, + "loss": 0.0048, + "step": 2350 + }, + { + "epoch": 1.2146718146718147, + "grad_norm": 0.09349490891133622, + "learning_rate": 9.68436800787461e-05, + "loss": 0.0048, + "step": 2360 + }, + { + "epoch": 1.2198198198198198, + "grad_norm": 0.08645959165196554, + "learning_rate": 9.679112003724433e-05, + "loss": 0.0047, + "step": 2370 + }, + { + "epoch": 1.224967824967825, + "grad_norm": 0.10081362401668935, + "learning_rate": 9.673814048189765e-05, + "loss": 0.0049, + "step": 2380 + }, + { + "epoch": 1.2301158301158301, + "grad_norm": 0.08712722386999017, + "learning_rate": 9.668474188770342e-05, + "loss": 0.0046, + "step": 2390 + }, + { + "epoch": 1.2352638352638352, + "grad_norm": 0.07793748236883287, + "learning_rate": 9.663092473341591e-05, + "loss": 0.0045, + "step": 2400 + }, + { + "epoch": 1.2404118404118405, + "grad_norm": 0.08419470824276815, + "learning_rate": 9.657668950154212e-05, + "loss": 0.0045, + "step": 2410 + }, + { + "epoch": 1.2455598455598456, + "grad_norm": 0.08910987303942121, + "learning_rate": 9.652203667833732e-05, + "loss": 0.0046, + "step": 2420 + }, + { + "epoch": 1.2507078507078506, + "grad_norm": 0.09000059511882977, + "learning_rate": 9.646696675380084e-05, + "loss": 0.0046, + "step": 2430 + }, + { + "epoch": 1.255855855855856, + "grad_norm": 0.08973547907365796, + "learning_rate": 9.641148022167157e-05, + "loss": 0.0045, + "step": 2440 + }, + { + "epoch": 1.261003861003861, + "grad_norm": 0.08752137232187301, + "learning_rate": 9.635557757942356e-05, + "loss": 0.0045, + "step": 2450 + }, + { + "epoch": 1.266151866151866, + "grad_norm": 0.08625521978145931, + "learning_rate": 9.629925932826158e-05, + "loss": 0.0044, + "step": 2460 + }, + { + "epoch": 1.2712998712998713, + "grad_norm": 0.09652129154021626, + "learning_rate": 9.62425259731166e-05, + "loss": 0.0044, + "step": 2470 + }, + { + "epoch": 1.2764478764478764, + "grad_norm": 0.07738037320281026, + "learning_rate": 9.61853780226413e-05, + "loss": 0.0043, + "step": 2480 + }, + { + "epoch": 1.2815958815958817, + "grad_norm": 0.08005866211189636, + "learning_rate": 9.612781598920545e-05, + "loss": 0.0042, + "step": 2490 + }, + { + "epoch": 1.2867438867438867, + "grad_norm": 0.07390360062985053, + "learning_rate": 9.60698403888914e-05, + "loss": 0.0043, + "step": 2500 + }, + { + "epoch": 1.291891891891892, + "grad_norm": 0.07290137906930576, + "learning_rate": 9.601145174148938e-05, + "loss": 0.0041, + "step": 2510 + }, + { + "epoch": 1.297039897039897, + "grad_norm": 0.08077140524538812, + "learning_rate": 9.595265057049284e-05, + "loss": 0.0043, + "step": 2520 + }, + { + "epoch": 1.3021879021879021, + "grad_norm": 0.071170385653506, + "learning_rate": 9.589343740309382e-05, + "loss": 0.0042, + "step": 2530 + }, + { + "epoch": 1.3073359073359074, + "grad_norm": 0.08200339249188611, + "learning_rate": 9.583381277017816e-05, + "loss": 0.004, + "step": 2540 + }, + { + "epoch": 1.3124839124839125, + "grad_norm": 0.07521720836461765, + "learning_rate": 9.577377720632079e-05, + "loss": 0.0039, + "step": 2550 + }, + { + "epoch": 1.3176319176319176, + "grad_norm": 0.0901292008046331, + "learning_rate": 9.57133312497809e-05, + "loss": 0.0041, + "step": 2560 + }, + { + "epoch": 1.3227799227799228, + "grad_norm": 0.08285967408155427, + "learning_rate": 9.56524754424971e-05, + "loss": 0.004, + "step": 2570 + }, + { + "epoch": 1.327927927927928, + "grad_norm": 0.09219709170644368, + "learning_rate": 9.55912103300826e-05, + "loss": 0.0041, + "step": 2580 + }, + { + "epoch": 1.333075933075933, + "grad_norm": 0.06662885941521572, + "learning_rate": 9.552953646182037e-05, + "loss": 0.004, + "step": 2590 + }, + { + "epoch": 1.3382239382239383, + "grad_norm": 0.08022484610511026, + "learning_rate": 9.546745439065805e-05, + "loss": 0.004, + "step": 2600 + }, + { + "epoch": 1.3433719433719433, + "grad_norm": 0.08810609300651148, + "learning_rate": 9.540496467320317e-05, + "loss": 0.004, + "step": 2610 + }, + { + "epoch": 1.3485199485199484, + "grad_norm": 0.08370070467094351, + "learning_rate": 9.534206786971802e-05, + "loss": 0.004, + "step": 2620 + }, + { + "epoch": 1.3536679536679537, + "grad_norm": 0.08571191252220195, + "learning_rate": 9.527876454411476e-05, + "loss": 0.0039, + "step": 2630 + }, + { + "epoch": 1.3588159588159587, + "grad_norm": 0.08575267541892966, + "learning_rate": 9.521505526395025e-05, + "loss": 0.0039, + "step": 2640 + }, + { + "epoch": 1.363963963963964, + "grad_norm": 0.0772459071410357, + "learning_rate": 9.515094060042098e-05, + "loss": 0.0039, + "step": 2650 + }, + { + "epoch": 1.369111969111969, + "grad_norm": 0.09154097043649315, + "learning_rate": 9.508642112835803e-05, + "loss": 0.0036, + "step": 2660 + }, + { + "epoch": 1.3742599742599744, + "grad_norm": 0.06979887713205823, + "learning_rate": 9.502149742622186e-05, + "loss": 0.0036, + "step": 2670 + }, + { + "epoch": 1.3794079794079794, + "grad_norm": 0.07524766626872072, + "learning_rate": 9.495617007609708e-05, + "loss": 0.0037, + "step": 2680 + }, + { + "epoch": 1.3845559845559845, + "grad_norm": 0.07741077734917863, + "learning_rate": 9.489043966368729e-05, + "loss": 0.0036, + "step": 2690 + }, + { + "epoch": 1.3897039897039898, + "grad_norm": 0.07777205972495332, + "learning_rate": 9.482430677830984e-05, + "loss": 0.0036, + "step": 2700 + }, + { + "epoch": 1.3948519948519948, + "grad_norm": 0.08027699605432952, + "learning_rate": 9.475777201289049e-05, + "loss": 0.0037, + "step": 2710 + }, + { + "epoch": 1.4, + "grad_norm": 0.08878147480054278, + "learning_rate": 9.469083596395813e-05, + "loss": 0.0037, + "step": 2720 + }, + { + "epoch": 1.4051480051480052, + "grad_norm": 0.0824428011807005, + "learning_rate": 9.462349923163947e-05, + "loss": 0.0037, + "step": 2730 + }, + { + "epoch": 1.4102960102960103, + "grad_norm": 0.10261444743241076, + "learning_rate": 9.455576241965357e-05, + "loss": 0.0035, + "step": 2740 + }, + { + "epoch": 1.4154440154440153, + "grad_norm": 0.08949325884555213, + "learning_rate": 9.448762613530644e-05, + "loss": 0.0036, + "step": 2750 + }, + { + "epoch": 1.4205920205920206, + "grad_norm": 0.0723289853593256, + "learning_rate": 9.44190909894857e-05, + "loss": 0.0036, + "step": 2760 + }, + { + "epoch": 1.4257400257400257, + "grad_norm": 0.08857410702799448, + "learning_rate": 9.435015759665502e-05, + "loss": 0.0034, + "step": 2770 + }, + { + "epoch": 1.430888030888031, + "grad_norm": 0.08063193397260371, + "learning_rate": 9.428082657484858e-05, + "loss": 0.0035, + "step": 2780 + }, + { + "epoch": 1.436036036036036, + "grad_norm": 0.07967833602585864, + "learning_rate": 9.421109854566562e-05, + "loss": 0.0036, + "step": 2790 + }, + { + "epoch": 1.4411840411840413, + "grad_norm": 0.08170135657562853, + "learning_rate": 9.414097413426477e-05, + "loss": 0.0036, + "step": 2800 + }, + { + "epoch": 1.4463320463320464, + "grad_norm": 0.06868763716374629, + "learning_rate": 9.407045396935852e-05, + "loss": 0.0033, + "step": 2810 + }, + { + "epoch": 1.4514800514800514, + "grad_norm": 0.07632059161778292, + "learning_rate": 9.399953868320757e-05, + "loss": 0.0033, + "step": 2820 + }, + { + "epoch": 1.4566280566280567, + "grad_norm": 0.07781026078770122, + "learning_rate": 9.392822891161513e-05, + "loss": 0.0032, + "step": 2830 + }, + { + "epoch": 1.4617760617760618, + "grad_norm": 0.07814855533275045, + "learning_rate": 9.38565252939212e-05, + "loss": 0.0033, + "step": 2840 + }, + { + "epoch": 1.4669240669240668, + "grad_norm": 0.07181980467452527, + "learning_rate": 9.378442847299696e-05, + "loss": 0.0034, + "step": 2850 + }, + { + "epoch": 1.4720720720720721, + "grad_norm": 0.07540071489782683, + "learning_rate": 9.371193909523885e-05, + "loss": 0.0034, + "step": 2860 + }, + { + "epoch": 1.4772200772200772, + "grad_norm": 0.07474893132191902, + "learning_rate": 9.363905781056288e-05, + "loss": 0.0032, + "step": 2870 + }, + { + "epoch": 1.4823680823680823, + "grad_norm": 0.06752409841169732, + "learning_rate": 9.356578527239876e-05, + "loss": 0.0033, + "step": 2880 + }, + { + "epoch": 1.4875160875160875, + "grad_norm": 0.07227853197497423, + "learning_rate": 9.349212213768403e-05, + "loss": 0.0033, + "step": 2890 + }, + { + "epoch": 1.4926640926640926, + "grad_norm": 0.08116684705266587, + "learning_rate": 9.341806906685824e-05, + "loss": 0.0033, + "step": 2900 + }, + { + "epoch": 1.4978120978120977, + "grad_norm": 0.07203072835662648, + "learning_rate": 9.334362672385692e-05, + "loss": 0.0032, + "step": 2910 + }, + { + "epoch": 1.502960102960103, + "grad_norm": 0.06997940354981029, + "learning_rate": 9.326879577610574e-05, + "loss": 0.003, + "step": 2920 + }, + { + "epoch": 1.5081081081081082, + "grad_norm": 0.07405993830075755, + "learning_rate": 9.319357689451443e-05, + "loss": 0.0031, + "step": 2930 + }, + { + "epoch": 1.5132561132561133, + "grad_norm": 0.06306898034549013, + "learning_rate": 9.311797075347084e-05, + "loss": 0.0032, + "step": 2940 + }, + { + "epoch": 1.5184041184041184, + "grad_norm": 0.0673006276719181, + "learning_rate": 9.304197803083486e-05, + "loss": 0.003, + "step": 2950 + }, + { + "epoch": 1.5235521235521237, + "grad_norm": 0.06910575947165132, + "learning_rate": 9.296559940793231e-05, + "loss": 0.003, + "step": 2960 + }, + { + "epoch": 1.5287001287001287, + "grad_norm": 0.07701591471399939, + "learning_rate": 9.288883556954888e-05, + "loss": 0.003, + "step": 2970 + }, + { + "epoch": 1.5338481338481338, + "grad_norm": 0.07543347242871117, + "learning_rate": 9.2811687203924e-05, + "loss": 0.003, + "step": 2980 + }, + { + "epoch": 1.538996138996139, + "grad_norm": 0.06367656554256501, + "learning_rate": 9.273415500274463e-05, + "loss": 0.0031, + "step": 2990 + }, + { + "epoch": 1.5441441441441441, + "grad_norm": 0.07035242882025344, + "learning_rate": 9.265623966113908e-05, + "loss": 0.003, + "step": 3000 + }, + { + "epoch": 1.5492921492921492, + "grad_norm": 0.07832135035294531, + "learning_rate": 9.257794187767072e-05, + "loss": 0.0031, + "step": 3010 + }, + { + "epoch": 1.5544401544401545, + "grad_norm": 0.06599561720521474, + "learning_rate": 9.249926235433187e-05, + "loss": 0.003, + "step": 3020 + }, + { + "epoch": 1.5595881595881596, + "grad_norm": 0.05267911081489989, + "learning_rate": 9.242020179653732e-05, + "loss": 0.0029, + "step": 3030 + }, + { + "epoch": 1.5647361647361646, + "grad_norm": 0.06704594471912081, + "learning_rate": 9.234076091311809e-05, + "loss": 0.003, + "step": 3040 + }, + { + "epoch": 1.56988416988417, + "grad_norm": 0.06460368185603607, + "learning_rate": 9.226094041631515e-05, + "loss": 0.0029, + "step": 3050 + }, + { + "epoch": 1.5750321750321752, + "grad_norm": 0.05979445278270312, + "learning_rate": 9.218074102177288e-05, + "loss": 0.003, + "step": 3060 + }, + { + "epoch": 1.58018018018018, + "grad_norm": 0.08016520116607714, + "learning_rate": 9.210016344853276e-05, + "loss": 0.0029, + "step": 3070 + }, + { + "epoch": 1.5853281853281853, + "grad_norm": 0.06802817286962771, + "learning_rate": 9.201920841902692e-05, + "loss": 0.003, + "step": 3080 + }, + { + "epoch": 1.5904761904761906, + "grad_norm": 0.06797365347482685, + "learning_rate": 9.193787665907161e-05, + "loss": 0.0029, + "step": 3090 + }, + { + "epoch": 1.5956241956241957, + "grad_norm": 0.07102105091489752, + "learning_rate": 9.185616889786073e-05, + "loss": 0.0029, + "step": 3100 + }, + { + "epoch": 1.6007722007722007, + "grad_norm": 0.07171846203450846, + "learning_rate": 9.177408586795927e-05, + "loss": 0.0029, + "step": 3110 + }, + { + "epoch": 1.605920205920206, + "grad_norm": 0.06055340864499565, + "learning_rate": 9.169162830529682e-05, + "loss": 0.0027, + "step": 3120 + }, + { + "epoch": 1.611068211068211, + "grad_norm": 0.07533819698090054, + "learning_rate": 9.16087969491608e-05, + "loss": 0.0028, + "step": 3130 + }, + { + "epoch": 1.6162162162162161, + "grad_norm": 0.06743711306850025, + "learning_rate": 9.152559254219003e-05, + "loss": 0.0027, + "step": 3140 + }, + { + "epoch": 1.6213642213642214, + "grad_norm": 0.06056537426381112, + "learning_rate": 9.144201583036791e-05, + "loss": 0.0027, + "step": 3150 + }, + { + "epoch": 1.6265122265122265, + "grad_norm": 0.06508273546989307, + "learning_rate": 9.135806756301591e-05, + "loss": 0.0027, + "step": 3160 + }, + { + "epoch": 1.6316602316602316, + "grad_norm": 0.06830147531666561, + "learning_rate": 9.127374849278662e-05, + "loss": 0.0028, + "step": 3170 + }, + { + "epoch": 1.6368082368082368, + "grad_norm": 0.06217191434234582, + "learning_rate": 9.118905937565722e-05, + "loss": 0.0028, + "step": 3180 + }, + { + "epoch": 1.641956241956242, + "grad_norm": 0.05780489801778332, + "learning_rate": 9.110400097092256e-05, + "loss": 0.0027, + "step": 3190 + }, + { + "epoch": 1.647104247104247, + "grad_norm": 0.06254550151471086, + "learning_rate": 9.101857404118843e-05, + "loss": 0.0026, + "step": 3200 + }, + { + "epoch": 1.6522522522522523, + "grad_norm": 0.05814178748214669, + "learning_rate": 9.093277935236466e-05, + "loss": 0.0026, + "step": 3210 + }, + { + "epoch": 1.6574002574002575, + "grad_norm": 0.16718357391133418, + "learning_rate": 9.08466176736583e-05, + "loss": 0.0028, + "step": 3220 + }, + { + "epoch": 1.6625482625482626, + "grad_norm": 2.396831293896128, + "learning_rate": 9.076008977756676e-05, + "loss": 0.0466, + "step": 3230 + }, + { + "epoch": 1.6676962676962677, + "grad_norm": 2.7472945663163135, + "learning_rate": 9.067319643987076e-05, + "loss": 0.0557, + "step": 3240 + }, + { + "epoch": 1.672844272844273, + "grad_norm": 0.26885148545801296, + "learning_rate": 9.05859384396275e-05, + "loss": 0.0304, + "step": 3250 + }, + { + "epoch": 1.677992277992278, + "grad_norm": 0.12409899016396063, + "learning_rate": 9.04983165591636e-05, + "loss": 0.0095, + "step": 3260 + }, + { + "epoch": 1.683140283140283, + "grad_norm": 0.09294953853472421, + "learning_rate": 9.041033158406812e-05, + "loss": 0.0055, + "step": 3270 + }, + { + "epoch": 1.6882882882882884, + "grad_norm": 0.09689553530835725, + "learning_rate": 9.032198430318549e-05, + "loss": 0.0042, + "step": 3280 + }, + { + "epoch": 1.6934362934362934, + "grad_norm": 0.09140013573445376, + "learning_rate": 9.023327550860851e-05, + "loss": 0.0038, + "step": 3290 + }, + { + "epoch": 1.6985842985842985, + "grad_norm": 0.08116155041824391, + "learning_rate": 9.014420599567113e-05, + "loss": 0.0034, + "step": 3300 + }, + { + "epoch": 1.7037323037323038, + "grad_norm": 0.0679507754540547, + "learning_rate": 9.00547765629414e-05, + "loss": 0.0032, + "step": 3310 + }, + { + "epoch": 1.7088803088803088, + "grad_norm": 0.058242905790430746, + "learning_rate": 8.996498801221431e-05, + "loss": 0.003, + "step": 3320 + }, + { + "epoch": 1.714028314028314, + "grad_norm": 0.07351637623119718, + "learning_rate": 8.98748411485046e-05, + "loss": 0.0029, + "step": 3330 + }, + { + "epoch": 1.7191763191763192, + "grad_norm": 0.06328328422520454, + "learning_rate": 8.978433678003946e-05, + "loss": 0.0029, + "step": 3340 + }, + { + "epoch": 1.7243243243243245, + "grad_norm": 0.07268810964129091, + "learning_rate": 8.969347571825142e-05, + "loss": 0.0029, + "step": 3350 + }, + { + "epoch": 1.7294723294723293, + "grad_norm": 0.05741572542909016, + "learning_rate": 8.960225877777095e-05, + "loss": 0.0029, + "step": 3360 + }, + { + "epoch": 1.7346203346203346, + "grad_norm": 0.06233872470439857, + "learning_rate": 8.951068677641928e-05, + "loss": 0.0027, + "step": 3370 + }, + { + "epoch": 1.7397683397683399, + "grad_norm": 0.06306044970907165, + "learning_rate": 8.94187605352009e-05, + "loss": 0.0027, + "step": 3380 + }, + { + "epoch": 1.744916344916345, + "grad_norm": 0.05866957972844069, + "learning_rate": 8.932648087829643e-05, + "loss": 0.0026, + "step": 3390 + }, + { + "epoch": 1.75006435006435, + "grad_norm": 0.06351564249061811, + "learning_rate": 8.923384863305495e-05, + "loss": 0.0027, + "step": 3400 + }, + { + "epoch": 1.7552123552123553, + "grad_norm": 0.06311005104924097, + "learning_rate": 8.914086462998686e-05, + "loss": 0.0026, + "step": 3410 + }, + { + "epoch": 1.7603603603603604, + "grad_norm": 0.07175111106462256, + "learning_rate": 8.904752970275624e-05, + "loss": 0.0026, + "step": 3420 + }, + { + "epoch": 1.7655083655083654, + "grad_norm": 0.06069156181724329, + "learning_rate": 8.895384468817345e-05, + "loss": 0.0025, + "step": 3430 + }, + { + "epoch": 1.7706563706563707, + "grad_norm": 0.06445989759572321, + "learning_rate": 8.885981042618762e-05, + "loss": 0.0025, + "step": 3440 + }, + { + "epoch": 1.7758043758043758, + "grad_norm": 0.06092468872567865, + "learning_rate": 8.876542775987912e-05, + "loss": 0.0024, + "step": 3450 + }, + { + "epoch": 1.7809523809523808, + "grad_norm": 0.0610543681951457, + "learning_rate": 8.867069753545202e-05, + "loss": 0.0024, + "step": 3460 + }, + { + "epoch": 1.7861003861003861, + "grad_norm": 0.057721016476337414, + "learning_rate": 8.857562060222644e-05, + "loss": 0.0023, + "step": 3470 + }, + { + "epoch": 1.7912483912483914, + "grad_norm": 0.06437257969879298, + "learning_rate": 8.848019781263099e-05, + "loss": 0.0025, + "step": 3480 + }, + { + "epoch": 1.7963963963963963, + "grad_norm": 0.06569126814994662, + "learning_rate": 8.838443002219514e-05, + "loss": 0.0023, + "step": 3490 + }, + { + "epoch": 1.8015444015444015, + "grad_norm": 0.06150895775850208, + "learning_rate": 8.82883180895415e-05, + "loss": 0.0023, + "step": 3500 + }, + { + "epoch": 1.8066924066924068, + "grad_norm": 0.0616857154906518, + "learning_rate": 8.819186287637817e-05, + "loss": 0.0023, + "step": 3510 + }, + { + "epoch": 1.811840411840412, + "grad_norm": 0.09278079283374936, + "learning_rate": 8.809506524749097e-05, + "loss": 0.0025, + "step": 3520 + }, + { + "epoch": 1.816988416988417, + "grad_norm": 0.05632880872010278, + "learning_rate": 8.799792607073567e-05, + "loss": 0.0024, + "step": 3530 + }, + { + "epoch": 1.8221364221364222, + "grad_norm": 0.06087327975528164, + "learning_rate": 8.790044621703033e-05, + "loss": 0.0024, + "step": 3540 + }, + { + "epoch": 1.8272844272844273, + "grad_norm": 0.05140811610791699, + "learning_rate": 8.780262656034739e-05, + "loss": 0.0023, + "step": 3550 + }, + { + "epoch": 1.8324324324324324, + "grad_norm": 0.06674966884600453, + "learning_rate": 8.770446797770577e-05, + "loss": 0.0023, + "step": 3560 + }, + { + "epoch": 1.8375804375804377, + "grad_norm": 0.05618543547388763, + "learning_rate": 8.760597134916318e-05, + "loss": 0.0022, + "step": 3570 + }, + { + "epoch": 1.8427284427284427, + "grad_norm": 0.05237285038394349, + "learning_rate": 8.750713755780811e-05, + "loss": 0.0022, + "step": 3580 + }, + { + "epoch": 1.8478764478764478, + "grad_norm": 0.058744035782622876, + "learning_rate": 8.740796748975191e-05, + "loss": 0.0021, + "step": 3590 + }, + { + "epoch": 1.853024453024453, + "grad_norm": 0.06251312062428276, + "learning_rate": 8.730846203412094e-05, + "loss": 0.0021, + "step": 3600 + }, + { + "epoch": 1.8581724581724581, + "grad_norm": 0.05660174173246188, + "learning_rate": 8.720862208304847e-05, + "loss": 0.0022, + "step": 3610 + }, + { + "epoch": 1.8633204633204632, + "grad_norm": 0.05460329788545314, + "learning_rate": 8.710844853166677e-05, + "loss": 0.0021, + "step": 3620 + }, + { + "epoch": 1.8684684684684685, + "grad_norm": 0.05920176351989063, + "learning_rate": 8.70079422780991e-05, + "loss": 0.0021, + "step": 3630 + }, + { + "epoch": 1.8736164736164738, + "grad_norm": 0.05975993421974977, + "learning_rate": 8.690710422345156e-05, + "loss": 0.0021, + "step": 3640 + }, + { + "epoch": 1.8787644787644786, + "grad_norm": 0.07261290977712913, + "learning_rate": 8.680593527180512e-05, + "loss": 0.0022, + "step": 3650 + }, + { + "epoch": 1.883912483912484, + "grad_norm": 0.057238529150264346, + "learning_rate": 8.67044363302074e-05, + "loss": 0.0021, + "step": 3660 + }, + { + "epoch": 1.8890604890604892, + "grad_norm": 0.05467721608387215, + "learning_rate": 8.660260830866467e-05, + "loss": 0.0021, + "step": 3670 + }, + { + "epoch": 1.8942084942084942, + "grad_norm": 0.05845129114063532, + "learning_rate": 8.650045212013358e-05, + "loss": 0.0023, + "step": 3680 + }, + { + "epoch": 1.8993564993564993, + "grad_norm": 0.07268624727169087, + "learning_rate": 8.639796868051305e-05, + "loss": 0.0021, + "step": 3690 + }, + { + "epoch": 1.9045045045045046, + "grad_norm": 0.05876373181946239, + "learning_rate": 8.629515890863596e-05, + "loss": 0.0022, + "step": 3700 + }, + { + "epoch": 1.9096525096525097, + "grad_norm": 0.048951820788902485, + "learning_rate": 8.619202372626109e-05, + "loss": 0.0021, + "step": 3710 + }, + { + "epoch": 1.9148005148005147, + "grad_norm": 0.05561005341623571, + "learning_rate": 8.608856405806464e-05, + "loss": 0.0019, + "step": 3720 + }, + { + "epoch": 1.91994851994852, + "grad_norm": 0.057100787317847096, + "learning_rate": 8.598478083163208e-05, + "loss": 0.002, + "step": 3730 + }, + { + "epoch": 1.925096525096525, + "grad_norm": 0.059660059973863006, + "learning_rate": 8.588067497744983e-05, + "loss": 0.0021, + "step": 3740 + }, + { + "epoch": 1.9302445302445301, + "grad_norm": 0.047168011131905614, + "learning_rate": 8.57762474288968e-05, + "loss": 0.0019, + "step": 3750 + }, + { + "epoch": 1.9353925353925354, + "grad_norm": 0.0594860132000797, + "learning_rate": 8.567149912223624e-05, + "loss": 0.002, + "step": 3760 + }, + { + "epoch": 1.9405405405405407, + "grad_norm": 0.060916594149851046, + "learning_rate": 8.556643099660706e-05, + "loss": 0.0019, + "step": 3770 + }, + { + "epoch": 1.9456885456885455, + "grad_norm": 0.06191170974205215, + "learning_rate": 8.546104399401567e-05, + "loss": 0.0019, + "step": 3780 + }, + { + "epoch": 1.9508365508365508, + "grad_norm": 0.059050634437477476, + "learning_rate": 8.535533905932738e-05, + "loss": 0.002, + "step": 3790 + }, + { + "epoch": 1.9559845559845561, + "grad_norm": 0.05874211825594844, + "learning_rate": 8.5249317140258e-05, + "loss": 0.002, + "step": 3800 + }, + { + "epoch": 1.9611325611325612, + "grad_norm": 0.051714056463086085, + "learning_rate": 8.514297918736528e-05, + "loss": 0.002, + "step": 3810 + }, + { + "epoch": 1.9662805662805662, + "grad_norm": 0.05669285044549558, + "learning_rate": 8.503632615404047e-05, + "loss": 0.0018, + "step": 3820 + }, + { + "epoch": 1.9714285714285715, + "grad_norm": 0.0527546216463499, + "learning_rate": 8.492935899649971e-05, + "loss": 0.0019, + "step": 3830 + }, + { + "epoch": 1.9765765765765766, + "grad_norm": 0.0517620866256957, + "learning_rate": 8.482207867377546e-05, + "loss": 0.0019, + "step": 3840 + }, + { + "epoch": 1.9817245817245817, + "grad_norm": 0.05504170017730614, + "learning_rate": 8.471448614770792e-05, + "loss": 0.0019, + "step": 3850 + }, + { + "epoch": 1.986872586872587, + "grad_norm": 0.0531051637907689, + "learning_rate": 8.460658238293644e-05, + "loss": 0.0019, + "step": 3860 + }, + { + "epoch": 1.992020592020592, + "grad_norm": 0.06878662684799065, + "learning_rate": 8.449836834689078e-05, + "loss": 0.0019, + "step": 3870 + }, + { + "epoch": 1.997168597168597, + "grad_norm": 0.07389086836198688, + "learning_rate": 8.438984500978253e-05, + "loss": 0.002, + "step": 3880 + }, + { + "epoch": 2.002059202059202, + "grad_norm": 0.04129522947698218, + "learning_rate": 8.428101334459634e-05, + "loss": 0.0018, + "step": 3890 + }, + { + "epoch": 2.007207207207207, + "grad_norm": 0.07916532613092706, + "learning_rate": 8.41718743270813e-05, + "loss": 0.0015, + "step": 3900 + }, + { + "epoch": 2.0123552123552124, + "grad_norm": 0.04733101682802457, + "learning_rate": 8.406242893574198e-05, + "loss": 0.0015, + "step": 3910 + }, + { + "epoch": 2.0175032175032177, + "grad_norm": 0.0417837325535251, + "learning_rate": 8.395267815182996e-05, + "loss": 0.0014, + "step": 3920 + }, + { + "epoch": 2.0226512226512225, + "grad_norm": 0.052954653259521055, + "learning_rate": 8.384262295933477e-05, + "loss": 0.0013, + "step": 3930 + }, + { + "epoch": 2.027799227799228, + "grad_norm": 0.04596870277989711, + "learning_rate": 8.373226434497518e-05, + "loss": 0.0014, + "step": 3940 + }, + { + "epoch": 2.032947232947233, + "grad_norm": 0.04826449367607423, + "learning_rate": 8.362160329819038e-05, + "loss": 0.0012, + "step": 3950 + }, + { + "epoch": 2.038095238095238, + "grad_norm": 0.059447596021510715, + "learning_rate": 8.3510640811131e-05, + "loss": 0.0013, + "step": 3960 + }, + { + "epoch": 2.0432432432432432, + "grad_norm": 0.04596769846728861, + "learning_rate": 8.339937787865036e-05, + "loss": 0.0013, + "step": 3970 + }, + { + "epoch": 2.0483912483912485, + "grad_norm": 0.044589049577602416, + "learning_rate": 8.328781549829543e-05, + "loss": 0.0014, + "step": 3980 + }, + { + "epoch": 2.0535392535392534, + "grad_norm": 0.04119681252531216, + "learning_rate": 8.317595467029795e-05, + "loss": 0.0014, + "step": 3990 + }, + { + "epoch": 2.0586872586872587, + "grad_norm": 0.05580855727608084, + "learning_rate": 8.306379639756544e-05, + "loss": 0.0012, + "step": 4000 + }, + { + "epoch": 2.063835263835264, + "grad_norm": 0.04940964639449694, + "learning_rate": 8.29513416856722e-05, + "loss": 0.0013, + "step": 4010 + }, + { + "epoch": 2.068983268983269, + "grad_norm": 0.04780393223229874, + "learning_rate": 8.283859154285032e-05, + "loss": 0.0013, + "step": 4020 + }, + { + "epoch": 2.074131274131274, + "grad_norm": 0.0581497942061673, + "learning_rate": 8.272554697998063e-05, + "loss": 0.0013, + "step": 4030 + }, + { + "epoch": 2.0792792792792794, + "grad_norm": 0.04439803542359854, + "learning_rate": 8.261220901058362e-05, + "loss": 0.0013, + "step": 4040 + }, + { + "epoch": 2.084427284427284, + "grad_norm": 0.03893925557971465, + "learning_rate": 8.249857865081037e-05, + "loss": 0.0013, + "step": 4050 + }, + { + "epoch": 2.0895752895752895, + "grad_norm": 0.05448252248762289, + "learning_rate": 8.238465691943344e-05, + "loss": 0.0014, + "step": 4060 + }, + { + "epoch": 2.0947232947232948, + "grad_norm": 0.04777291394045677, + "learning_rate": 8.227044483783772e-05, + "loss": 0.0013, + "step": 4070 + }, + { + "epoch": 2.0998712998713, + "grad_norm": 0.04531698982390845, + "learning_rate": 8.215594343001133e-05, + "loss": 0.0014, + "step": 4080 + }, + { + "epoch": 2.105019305019305, + "grad_norm": 0.05289715107833794, + "learning_rate": 8.204115372253635e-05, + "loss": 0.0014, + "step": 4090 + }, + { + "epoch": 2.11016731016731, + "grad_norm": 0.05000518964787425, + "learning_rate": 8.192607674457966e-05, + "loss": 0.0014, + "step": 4100 + }, + { + "epoch": 2.1153153153153155, + "grad_norm": 0.07423565873105893, + "learning_rate": 8.181071352788371e-05, + "loss": 0.0013, + "step": 4110 + }, + { + "epoch": 2.1204633204633203, + "grad_norm": 0.04387669724085158, + "learning_rate": 8.169506510675733e-05, + "loss": 0.0013, + "step": 4120 + }, + { + "epoch": 2.1256113256113256, + "grad_norm": 0.054902691601927234, + "learning_rate": 8.15791325180663e-05, + "loss": 0.0014, + "step": 4130 + }, + { + "epoch": 2.130759330759331, + "grad_norm": 0.048346239723777266, + "learning_rate": 8.146291680122427e-05, + "loss": 0.0014, + "step": 4140 + }, + { + "epoch": 2.1359073359073357, + "grad_norm": 0.05113058555459646, + "learning_rate": 8.13464189981832e-05, + "loss": 0.0012, + "step": 4150 + }, + { + "epoch": 2.141055341055341, + "grad_norm": 0.04692623057151712, + "learning_rate": 8.122964015342427e-05, + "loss": 0.0013, + "step": 4160 + }, + { + "epoch": 2.1462033462033463, + "grad_norm": 0.049807948136136, + "learning_rate": 8.111258131394828e-05, + "loss": 0.0013, + "step": 4170 + }, + { + "epoch": 2.1513513513513516, + "grad_norm": 0.04810393599315422, + "learning_rate": 8.099524352926643e-05, + "loss": 0.0013, + "step": 4180 + }, + { + "epoch": 2.1564993564993564, + "grad_norm": 0.04804529263899986, + "learning_rate": 8.087762785139084e-05, + "loss": 0.0014, + "step": 4190 + }, + { + "epoch": 2.1616473616473617, + "grad_norm": 0.04394140785220738, + "learning_rate": 8.07597353348251e-05, + "loss": 0.0012, + "step": 4200 + }, + { + "epoch": 2.166795366795367, + "grad_norm": 0.04386355702357804, + "learning_rate": 8.064156703655489e-05, + "loss": 0.0011, + "step": 4210 + }, + { + "epoch": 2.171943371943372, + "grad_norm": 0.04119947072133049, + "learning_rate": 8.052312401603847e-05, + "loss": 0.0012, + "step": 4220 + }, + { + "epoch": 2.177091377091377, + "grad_norm": 0.04030666528465274, + "learning_rate": 8.040440733519709e-05, + "loss": 0.0012, + "step": 4230 + }, + { + "epoch": 2.1822393822393824, + "grad_norm": 0.057812767601739, + "learning_rate": 8.028541805840561e-05, + "loss": 0.0012, + "step": 4240 + }, + { + "epoch": 2.1873873873873872, + "grad_norm": 0.047874075229963925, + "learning_rate": 8.016615725248289e-05, + "loss": 0.0015, + "step": 4250 + }, + { + "epoch": 2.1925353925353925, + "grad_norm": 0.059625625401084195, + "learning_rate": 8.00466259866822e-05, + "loss": 0.0014, + "step": 4260 + }, + { + "epoch": 2.197683397683398, + "grad_norm": 0.050376692026641155, + "learning_rate": 7.992682533268175e-05, + "loss": 0.0013, + "step": 4270 + }, + { + "epoch": 2.2028314028314027, + "grad_norm": 0.048462338863960994, + "learning_rate": 7.980675636457487e-05, + "loss": 0.0014, + "step": 4280 + }, + { + "epoch": 2.207979407979408, + "grad_norm": 0.04835528405206845, + "learning_rate": 7.968642015886058e-05, + "loss": 0.0013, + "step": 4290 + }, + { + "epoch": 2.2131274131274132, + "grad_norm": 0.039780282771479415, + "learning_rate": 7.95658177944339e-05, + "loss": 0.0011, + "step": 4300 + }, + { + "epoch": 2.218275418275418, + "grad_norm": 0.04452063196388458, + "learning_rate": 7.944495035257605e-05, + "loss": 0.0012, + "step": 4310 + }, + { + "epoch": 2.2234234234234234, + "grad_norm": 0.0456971239108907, + "learning_rate": 7.932381891694493e-05, + "loss": 0.0012, + "step": 4320 + }, + { + "epoch": 2.2285714285714286, + "grad_norm": 0.04293091701214276, + "learning_rate": 7.920242457356526e-05, + "loss": 0.0012, + "step": 4330 + }, + { + "epoch": 2.2337194337194335, + "grad_norm": 0.04093195000280922, + "learning_rate": 7.908076841081895e-05, + "loss": 0.0013, + "step": 4340 + }, + { + "epoch": 2.2388674388674388, + "grad_norm": 0.0424664722931835, + "learning_rate": 7.895885151943527e-05, + "loss": 0.0012, + "step": 4350 + }, + { + "epoch": 2.244015444015444, + "grad_norm": 0.0821934383463096, + "learning_rate": 7.883667499248111e-05, + "loss": 0.0013, + "step": 4360 + }, + { + "epoch": 2.2491634491634493, + "grad_norm": 0.039405424949199575, + "learning_rate": 7.871423992535116e-05, + "loss": 0.0012, + "step": 4370 + }, + { + "epoch": 2.254311454311454, + "grad_norm": 0.04228642502788665, + "learning_rate": 7.859154741575809e-05, + "loss": 0.0012, + "step": 4380 + }, + { + "epoch": 2.2594594594594595, + "grad_norm": 0.04986484606810717, + "learning_rate": 7.846859856372275e-05, + "loss": 0.0012, + "step": 4390 + }, + { + "epoch": 2.2646074646074648, + "grad_norm": 0.06381252118101394, + "learning_rate": 7.834539447156424e-05, + "loss": 0.0013, + "step": 4400 + }, + { + "epoch": 2.2697554697554696, + "grad_norm": 0.0476340369024287, + "learning_rate": 7.822193624389005e-05, + "loss": 0.0013, + "step": 4410 + }, + { + "epoch": 2.274903474903475, + "grad_norm": 0.053280169836871914, + "learning_rate": 7.809822498758618e-05, + "loss": 0.0013, + "step": 4420 + }, + { + "epoch": 2.28005148005148, + "grad_norm": 0.0471759251214204, + "learning_rate": 7.797426181180723e-05, + "loss": 0.0011, + "step": 4430 + }, + { + "epoch": 2.285199485199485, + "grad_norm": 0.05750087294114376, + "learning_rate": 7.785004782796635e-05, + "loss": 0.0012, + "step": 4440 + }, + { + "epoch": 2.2903474903474903, + "grad_norm": 0.04433322895287332, + "learning_rate": 7.772558414972546e-05, + "loss": 0.0012, + "step": 4450 + }, + { + "epoch": 2.2954954954954956, + "grad_norm": 0.05145684185533547, + "learning_rate": 7.760087189298509e-05, + "loss": 0.0012, + "step": 4460 + }, + { + "epoch": 2.300643500643501, + "grad_norm": 0.05085435178874306, + "learning_rate": 7.747591217587443e-05, + "loss": 0.0012, + "step": 4470 + }, + { + "epoch": 2.3057915057915057, + "grad_norm": 0.05084038319068861, + "learning_rate": 7.735070611874137e-05, + "loss": 0.0013, + "step": 4480 + }, + { + "epoch": 2.310939510939511, + "grad_norm": 0.046795622343555554, + "learning_rate": 7.722525484414236e-05, + "loss": 0.0013, + "step": 4490 + }, + { + "epoch": 2.3160875160875163, + "grad_norm": 0.043906153913954984, + "learning_rate": 7.70995594768324e-05, + "loss": 0.0014, + "step": 4500 + }, + { + "epoch": 2.321235521235521, + "grad_norm": 0.059773455723487405, + "learning_rate": 7.6973621143755e-05, + "loss": 0.0013, + "step": 4510 + }, + { + "epoch": 2.3263835263835264, + "grad_norm": 0.05604359193591503, + "learning_rate": 7.684744097403191e-05, + "loss": 0.0012, + "step": 4520 + }, + { + "epoch": 2.3315315315315317, + "grad_norm": 0.04675320396248994, + "learning_rate": 7.672102009895321e-05, + "loss": 0.0013, + "step": 4530 + }, + { + "epoch": 2.3366795366795365, + "grad_norm": 0.04509813399138995, + "learning_rate": 7.659435965196701e-05, + "loss": 0.0013, + "step": 4540 + }, + { + "epoch": 2.341827541827542, + "grad_norm": 0.0576880617509529, + "learning_rate": 7.646746076866939e-05, + "loss": 0.0014, + "step": 4550 + }, + { + "epoch": 2.346975546975547, + "grad_norm": 0.06321893688694276, + "learning_rate": 7.634032458679412e-05, + "loss": 0.0018, + "step": 4560 + }, + { + "epoch": 2.352123552123552, + "grad_norm": 0.1633121647089786, + "learning_rate": 7.621295224620253e-05, + "loss": 0.0035, + "step": 4570 + }, + { + "epoch": 2.3572715572715572, + "grad_norm": 0.08921315729148695, + "learning_rate": 7.608534488887327e-05, + "loss": 0.0022, + "step": 4580 + }, + { + "epoch": 2.3624195624195625, + "grad_norm": 0.057771731238916685, + "learning_rate": 7.595750365889209e-05, + "loss": 0.0019, + "step": 4590 + }, + { + "epoch": 2.3675675675675674, + "grad_norm": 0.046773705729556504, + "learning_rate": 7.582942970244155e-05, + "loss": 0.0016, + "step": 4600 + }, + { + "epoch": 2.3727155727155727, + "grad_norm": 0.061357062741325005, + "learning_rate": 7.570112416779078e-05, + "loss": 0.0016, + "step": 4610 + }, + { + "epoch": 2.377863577863578, + "grad_norm": 0.06208953521085493, + "learning_rate": 7.557258820528514e-05, + "loss": 0.0015, + "step": 4620 + }, + { + "epoch": 2.383011583011583, + "grad_norm": 0.05563724202044124, + "learning_rate": 7.544382296733592e-05, + "loss": 0.0014, + "step": 4630 + }, + { + "epoch": 2.388159588159588, + "grad_norm": 0.054185493414282304, + "learning_rate": 7.531482960841004e-05, + "loss": 0.0013, + "step": 4640 + }, + { + "epoch": 2.3933075933075934, + "grad_norm": 0.04752869142758457, + "learning_rate": 7.518560928501969e-05, + "loss": 0.0013, + "step": 4650 + }, + { + "epoch": 2.3984555984555986, + "grad_norm": 0.05632456808738891, + "learning_rate": 7.505616315571192e-05, + "loss": 0.0014, + "step": 4660 + }, + { + "epoch": 2.4036036036036035, + "grad_norm": 0.04523988270062936, + "learning_rate": 7.492649238105828e-05, + "loss": 0.0012, + "step": 4670 + }, + { + "epoch": 2.4087516087516088, + "grad_norm": 0.04001437246775281, + "learning_rate": 7.479659812364448e-05, + "loss": 0.0012, + "step": 4680 + }, + { + "epoch": 2.413899613899614, + "grad_norm": 0.04689498867230884, + "learning_rate": 7.466648154805982e-05, + "loss": 0.0011, + "step": 4690 + }, + { + "epoch": 2.419047619047619, + "grad_norm": 0.04711399995411257, + "learning_rate": 7.453614382088687e-05, + "loss": 0.0012, + "step": 4700 + }, + { + "epoch": 2.424195624195624, + "grad_norm": 0.050401789341980335, + "learning_rate": 7.440558611069095e-05, + "loss": 0.0012, + "step": 4710 + }, + { + "epoch": 2.4293436293436295, + "grad_norm": 0.04507054029023686, + "learning_rate": 7.427480958800974e-05, + "loss": 0.0013, + "step": 4720 + }, + { + "epoch": 2.4344916344916343, + "grad_norm": 0.05114129043648352, + "learning_rate": 7.414381542534266e-05, + "loss": 0.0013, + "step": 4730 + }, + { + "epoch": 2.4396396396396396, + "grad_norm": 0.050858156890527396, + "learning_rate": 7.401260479714041e-05, + "loss": 0.0012, + "step": 4740 + }, + { + "epoch": 2.444787644787645, + "grad_norm": 0.046113564407026235, + "learning_rate": 7.388117887979449e-05, + "loss": 0.0011, + "step": 4750 + }, + { + "epoch": 2.44993564993565, + "grad_norm": 0.04014385370106455, + "learning_rate": 7.37495388516266e-05, + "loss": 0.001, + "step": 4760 + }, + { + "epoch": 2.455083655083655, + "grad_norm": 0.04803165364348251, + "learning_rate": 7.361768589287808e-05, + "loss": 0.0011, + "step": 4770 + }, + { + "epoch": 2.4602316602316603, + "grad_norm": 0.033866842151233145, + "learning_rate": 7.348562118569935e-05, + "loss": 0.0011, + "step": 4780 + }, + { + "epoch": 2.4653796653796656, + "grad_norm": 0.051903863558592306, + "learning_rate": 7.335334591413925e-05, + "loss": 0.0011, + "step": 4790 + }, + { + "epoch": 2.4705276705276704, + "grad_norm": 0.03814836987756061, + "learning_rate": 7.322086126413455e-05, + "loss": 0.0011, + "step": 4800 + }, + { + "epoch": 2.4756756756756757, + "grad_norm": 0.039269243726615485, + "learning_rate": 7.308816842349918e-05, + "loss": 0.001, + "step": 4810 + }, + { + "epoch": 2.480823680823681, + "grad_norm": 0.04765524329024219, + "learning_rate": 7.295526858191362e-05, + "loss": 0.0011, + "step": 4820 + }, + { + "epoch": 2.485971685971686, + "grad_norm": 0.04951778818784927, + "learning_rate": 7.282216293091434e-05, + "loss": 0.0012, + "step": 4830 + }, + { + "epoch": 2.491119691119691, + "grad_norm": 0.04536576824454616, + "learning_rate": 7.268885266388296e-05, + "loss": 0.0011, + "step": 4840 + }, + { + "epoch": 2.4962676962676964, + "grad_norm": 0.0515320102284534, + "learning_rate": 7.255533897603562e-05, + "loss": 0.001, + "step": 4850 + }, + { + "epoch": 2.5014157014157012, + "grad_norm": 0.04600110428117649, + "learning_rate": 7.242162306441231e-05, + "loss": 0.001, + "step": 4860 + }, + { + "epoch": 2.5065637065637065, + "grad_norm": 0.041692098130997376, + "learning_rate": 7.228770612786604e-05, + "loss": 0.0011, + "step": 4870 + }, + { + "epoch": 2.511711711711712, + "grad_norm": 0.04488982334958329, + "learning_rate": 7.215358936705219e-05, + "loss": 0.001, + "step": 4880 + }, + { + "epoch": 2.5168597168597167, + "grad_norm": 0.048039789785818715, + "learning_rate": 7.201927398441765e-05, + "loss": 0.0011, + "step": 4890 + }, + { + "epoch": 2.522007722007722, + "grad_norm": 0.04538093295084992, + "learning_rate": 7.18847611841901e-05, + "loss": 0.001, + "step": 4900 + }, + { + "epoch": 2.5271557271557272, + "grad_norm": 0.04709836872218103, + "learning_rate": 7.175005217236728e-05, + "loss": 0.001, + "step": 4910 + }, + { + "epoch": 2.532303732303732, + "grad_norm": 0.04505447857446204, + "learning_rate": 7.161514815670598e-05, + "loss": 0.0011, + "step": 4920 + }, + { + "epoch": 2.5374517374517374, + "grad_norm": 0.05828384125867241, + "learning_rate": 7.148005034671142e-05, + "loss": 0.0011, + "step": 4930 + }, + { + "epoch": 2.5425997425997426, + "grad_norm": 0.046343388951512784, + "learning_rate": 7.134475995362623e-05, + "loss": 0.001, + "step": 4940 + }, + { + "epoch": 2.5477477477477475, + "grad_norm": 0.04707163071731421, + "learning_rate": 7.120927819041979e-05, + "loss": 0.001, + "step": 4950 + }, + { + "epoch": 2.5528957528957528, + "grad_norm": 0.04306173241714982, + "learning_rate": 7.107360627177714e-05, + "loss": 0.001, + "step": 4960 + }, + { + "epoch": 2.558043758043758, + "grad_norm": 0.044851977335899865, + "learning_rate": 7.093774541408823e-05, + "loss": 0.0011, + "step": 4970 + }, + { + "epoch": 2.5631917631917633, + "grad_norm": 0.040963399467547255, + "learning_rate": 7.080169683543701e-05, + "loss": 0.0011, + "step": 4980 + }, + { + "epoch": 2.568339768339768, + "grad_norm": 0.0524316895531215, + "learning_rate": 7.066546175559042e-05, + "loss": 0.001, + "step": 4990 + }, + { + "epoch": 2.5734877734877735, + "grad_norm": 0.04091367424148807, + "learning_rate": 7.052904139598752e-05, + "loss": 0.001, + "step": 5000 + }, + { + "epoch": 2.5786357786357788, + "grad_norm": 0.05066184832309622, + "learning_rate": 7.039243697972857e-05, + "loss": 0.0011, + "step": 5010 + }, + { + "epoch": 2.583783783783784, + "grad_norm": 0.04236548105650226, + "learning_rate": 7.025564973156396e-05, + "loss": 0.0009, + "step": 5020 + }, + { + "epoch": 2.588931788931789, + "grad_norm": 0.04623210387140029, + "learning_rate": 7.011868087788333e-05, + "loss": 0.001, + "step": 5030 + }, + { + "epoch": 2.594079794079794, + "grad_norm": 0.03783743894851547, + "learning_rate": 6.99815316467045e-05, + "loss": 0.0011, + "step": 5040 + }, + { + "epoch": 2.5992277992277995, + "grad_norm": 0.042867495704172615, + "learning_rate": 6.984420326766257e-05, + "loss": 0.0011, + "step": 5050 + }, + { + "epoch": 2.6043758043758043, + "grad_norm": 0.042286428164349944, + "learning_rate": 6.970669697199876e-05, + "loss": 0.0009, + "step": 5060 + }, + { + "epoch": 2.6095238095238096, + "grad_norm": 0.06287895185713929, + "learning_rate": 6.956901399254943e-05, + "loss": 0.001, + "step": 5070 + }, + { + "epoch": 2.614671814671815, + "grad_norm": 0.055451825887909506, + "learning_rate": 6.943115556373502e-05, + "loss": 0.001, + "step": 5080 + }, + { + "epoch": 2.6198198198198197, + "grad_norm": 0.037650966668449, + "learning_rate": 6.929312292154908e-05, + "loss": 0.0011, + "step": 5090 + }, + { + "epoch": 2.624967824967825, + "grad_norm": 0.0422236192284774, + "learning_rate": 6.915491730354697e-05, + "loss": 0.001, + "step": 5100 + }, + { + "epoch": 2.6301158301158303, + "grad_norm": 0.04148234941603724, + "learning_rate": 6.9016539948835e-05, + "loss": 0.001, + "step": 5110 + }, + { + "epoch": 2.635263835263835, + "grad_norm": 0.05471853033140722, + "learning_rate": 6.887799209805917e-05, + "loss": 0.001, + "step": 5120 + }, + { + "epoch": 2.6404118404118404, + "grad_norm": 0.04759262188474491, + "learning_rate": 6.873927499339407e-05, + "loss": 0.0011, + "step": 5130 + }, + { + "epoch": 2.6455598455598457, + "grad_norm": 0.043132316263661374, + "learning_rate": 6.860038987853182e-05, + "loss": 0.001, + "step": 5140 + }, + { + "epoch": 2.6507078507078505, + "grad_norm": 0.045288505100104404, + "learning_rate": 6.846133799867082e-05, + "loss": 0.0011, + "step": 5150 + }, + { + "epoch": 2.655855855855856, + "grad_norm": 0.03982061447257109, + "learning_rate": 6.832212060050465e-05, + "loss": 0.001, + "step": 5160 + }, + { + "epoch": 2.661003861003861, + "grad_norm": 0.03191140789524671, + "learning_rate": 6.818273893221088e-05, + "loss": 0.0009, + "step": 5170 + }, + { + "epoch": 2.666151866151866, + "grad_norm": 0.04219223914359278, + "learning_rate": 6.804319424343984e-05, + "loss": 0.001, + "step": 5180 + }, + { + "epoch": 2.6712998712998712, + "grad_norm": 0.04261141040671573, + "learning_rate": 6.790348778530347e-05, + "loss": 0.001, + "step": 5190 + }, + { + "epoch": 2.6764478764478765, + "grad_norm": 0.046770608070064175, + "learning_rate": 6.77636208103641e-05, + "loss": 0.0009, + "step": 5200 + }, + { + "epoch": 2.6815958815958814, + "grad_norm": 0.05046470198461108, + "learning_rate": 6.762359457262318e-05, + "loss": 0.0009, + "step": 5210 + }, + { + "epoch": 2.6867438867438866, + "grad_norm": 0.04578382268098655, + "learning_rate": 6.748341032751004e-05, + "loss": 0.0011, + "step": 5220 + }, + { + "epoch": 2.691891891891892, + "grad_norm": 0.03916013043164779, + "learning_rate": 6.734306933187068e-05, + "loss": 0.001, + "step": 5230 + }, + { + "epoch": 2.6970398970398968, + "grad_norm": 0.0495431816414551, + "learning_rate": 6.720257284395649e-05, + "loss": 0.0009, + "step": 5240 + }, + { + "epoch": 2.702187902187902, + "grad_norm": 0.028887936049044396, + "learning_rate": 6.706192212341287e-05, + "loss": 0.0008, + "step": 5250 + }, + { + "epoch": 2.7073359073359073, + "grad_norm": 0.04096988584464123, + "learning_rate": 6.692111843126812e-05, + "loss": 0.001, + "step": 5260 + }, + { + "epoch": 2.7124839124839126, + "grad_norm": 0.06842570554905965, + "learning_rate": 6.6780163029922e-05, + "loss": 0.001, + "step": 5270 + }, + { + "epoch": 2.7176319176319175, + "grad_norm": 0.0417571647859502, + "learning_rate": 6.66390571831344e-05, + "loss": 0.001, + "step": 5280 + }, + { + "epoch": 2.7227799227799228, + "grad_norm": 0.03695050716607467, + "learning_rate": 6.649780215601411e-05, + "loss": 0.001, + "step": 5290 + }, + { + "epoch": 2.727927927927928, + "grad_norm": 0.0391813604064568, + "learning_rate": 6.635639921500742e-05, + "loss": 0.0009, + "step": 5300 + }, + { + "epoch": 2.7330759330759333, + "grad_norm": 0.04228547410069994, + "learning_rate": 6.621484962788671e-05, + "loss": 0.0011, + "step": 5310 + }, + { + "epoch": 2.738223938223938, + "grad_norm": 0.043524209122763294, + "learning_rate": 6.607315466373923e-05, + "loss": 0.0009, + "step": 5320 + }, + { + "epoch": 2.7433719433719435, + "grad_norm": 0.05219296497463659, + "learning_rate": 6.593131559295551e-05, + "loss": 0.0009, + "step": 5330 + }, + { + "epoch": 2.7485199485199487, + "grad_norm": 0.04065132149131985, + "learning_rate": 6.578933368721824e-05, + "loss": 0.001, + "step": 5340 + }, + { + "epoch": 2.7536679536679536, + "grad_norm": 0.040021905728438036, + "learning_rate": 6.564721021949061e-05, + "loss": 0.001, + "step": 5350 + }, + { + "epoch": 2.758815958815959, + "grad_norm": 0.03991881038489872, + "learning_rate": 6.550494646400506e-05, + "loss": 0.0008, + "step": 5360 + }, + { + "epoch": 2.763963963963964, + "grad_norm": 0.0428200375670312, + "learning_rate": 6.536254369625176e-05, + "loss": 0.0008, + "step": 5370 + }, + { + "epoch": 2.769111969111969, + "grad_norm": 0.040946393497408425, + "learning_rate": 6.52200031929673e-05, + "loss": 0.0008, + "step": 5380 + }, + { + "epoch": 2.7742599742599743, + "grad_norm": 0.043599458814907815, + "learning_rate": 6.507732623212308e-05, + "loss": 0.0008, + "step": 5390 + }, + { + "epoch": 2.7794079794079796, + "grad_norm": 0.04744156312077381, + "learning_rate": 6.493451409291394e-05, + "loss": 0.0009, + "step": 5400 + }, + { + "epoch": 2.7845559845559844, + "grad_norm": 0.033867584735995535, + "learning_rate": 6.479156805574675e-05, + "loss": 0.0009, + "step": 5410 + }, + { + "epoch": 2.7897039897039897, + "grad_norm": 0.05131657185596086, + "learning_rate": 6.464848940222882e-05, + "loss": 0.0009, + "step": 5420 + }, + { + "epoch": 2.794851994851995, + "grad_norm": 0.04148534948819875, + "learning_rate": 6.450527941515642e-05, + "loss": 0.0008, + "step": 5430 + }, + { + "epoch": 2.8, + "grad_norm": 0.04655515368380606, + "learning_rate": 6.43619393785034e-05, + "loss": 0.0009, + "step": 5440 + }, + { + "epoch": 2.805148005148005, + "grad_norm": 0.03710387814506504, + "learning_rate": 6.421847057740952e-05, + "loss": 0.0009, + "step": 5450 + }, + { + "epoch": 2.8102960102960104, + "grad_norm": 0.03913002237594932, + "learning_rate": 6.407487429816902e-05, + "loss": 0.001, + "step": 5460 + }, + { + "epoch": 2.8154440154440152, + "grad_norm": 0.03584996106534604, + "learning_rate": 6.393115182821907e-05, + "loss": 0.0009, + "step": 5470 + }, + { + "epoch": 2.8205920205920205, + "grad_norm": 0.0417667097051815, + "learning_rate": 6.378730445612825e-05, + "loss": 0.0009, + "step": 5480 + }, + { + "epoch": 2.825740025740026, + "grad_norm": 0.03784326586147908, + "learning_rate": 6.364333347158493e-05, + "loss": 0.0009, + "step": 5490 + }, + { + "epoch": 2.8308880308880306, + "grad_norm": 0.04164280280096474, + "learning_rate": 6.349924016538576e-05, + "loss": 0.0009, + "step": 5500 + }, + { + "epoch": 2.836036036036036, + "grad_norm": 0.048700348942891436, + "learning_rate": 6.335502582942408e-05, + "loss": 0.0009, + "step": 5510 + }, + { + "epoch": 2.841184041184041, + "grad_norm": 0.047092943022912524, + "learning_rate": 6.321069175667841e-05, + "loss": 0.0009, + "step": 5520 + }, + { + "epoch": 2.846332046332046, + "grad_norm": 0.053992305467123365, + "learning_rate": 6.306623924120068e-05, + "loss": 0.0009, + "step": 5530 + }, + { + "epoch": 2.8514800514800513, + "grad_norm": 0.050364752497194444, + "learning_rate": 6.292166957810482e-05, + "loss": 0.001, + "step": 5540 + }, + { + "epoch": 2.8566280566280566, + "grad_norm": 0.04054969403664085, + "learning_rate": 6.277698406355503e-05, + "loss": 0.0008, + "step": 5550 + }, + { + "epoch": 2.861776061776062, + "grad_norm": 0.04324075746627249, + "learning_rate": 6.263218399475423e-05, + "loss": 0.0009, + "step": 5560 + }, + { + "epoch": 2.8669240669240668, + "grad_norm": 0.04190148686351726, + "learning_rate": 6.248727066993234e-05, + "loss": 0.0008, + "step": 5570 + }, + { + "epoch": 2.872072072072072, + "grad_norm": 0.04758273172777981, + "learning_rate": 6.234224538833475e-05, + "loss": 0.0009, + "step": 5580 + }, + { + "epoch": 2.8772200772200773, + "grad_norm": 0.03685424905404639, + "learning_rate": 6.21971094502106e-05, + "loss": 0.0009, + "step": 5590 + }, + { + "epoch": 2.8823680823680826, + "grad_norm": 0.05051965251778674, + "learning_rate": 6.205186415680114e-05, + "loss": 0.0009, + "step": 5600 + }, + { + "epoch": 2.8875160875160875, + "grad_norm": 0.04104162055687117, + "learning_rate": 6.190651081032805e-05, + "loss": 0.0008, + "step": 5610 + }, + { + "epoch": 2.8926640926640927, + "grad_norm": 0.03771500173197802, + "learning_rate": 6.17610507139818e-05, + "loss": 0.0009, + "step": 5620 + }, + { + "epoch": 2.897812097812098, + "grad_norm": 0.03784406844412519, + "learning_rate": 6.161548517190992e-05, + "loss": 0.0008, + "step": 5630 + }, + { + "epoch": 2.902960102960103, + "grad_norm": 0.03472842327768882, + "learning_rate": 6.146981548920536e-05, + "loss": 0.0008, + "step": 5640 + }, + { + "epoch": 2.908108108108108, + "grad_norm": 0.038697895799216266, + "learning_rate": 6.132404297189473e-05, + "loss": 0.0008, + "step": 5650 + }, + { + "epoch": 2.9132561132561134, + "grad_norm": 0.038331027361052525, + "learning_rate": 6.117816892692666e-05, + "loss": 0.0008, + "step": 5660 + }, + { + "epoch": 2.9184041184041183, + "grad_norm": 0.06664634073497308, + "learning_rate": 6.1032194662160036e-05, + "loss": 0.0009, + "step": 5670 + }, + { + "epoch": 2.9235521235521236, + "grad_norm": 0.04535119884086214, + "learning_rate": 6.088612148635222e-05, + "loss": 0.0008, + "step": 5680 + }, + { + "epoch": 2.928700128700129, + "grad_norm": 0.03436794949873355, + "learning_rate": 6.073995070914744e-05, + "loss": 0.0009, + "step": 5690 + }, + { + "epoch": 2.9338481338481337, + "grad_norm": 0.0378410295888514, + "learning_rate": 6.0593683641065017e-05, + "loss": 0.0008, + "step": 5700 + }, + { + "epoch": 2.938996138996139, + "grad_norm": 0.04860576998481387, + "learning_rate": 6.0447321593487496e-05, + "loss": 0.0008, + "step": 5710 + }, + { + "epoch": 2.9441441441441443, + "grad_norm": 0.040173916389819236, + "learning_rate": 6.0300865878649016e-05, + "loss": 0.0008, + "step": 5720 + }, + { + "epoch": 2.949292149292149, + "grad_norm": 0.03839135507716829, + "learning_rate": 6.015431780962353e-05, + "loss": 0.0008, + "step": 5730 + }, + { + "epoch": 2.9544401544401544, + "grad_norm": 0.04577373007241446, + "learning_rate": 6.000767870031299e-05, + "loss": 0.0008, + "step": 5740 + }, + { + "epoch": 2.9595881595881597, + "grad_norm": 0.03058337645850844, + "learning_rate": 5.986094986543558e-05, + "loss": 0.0008, + "step": 5750 + }, + { + "epoch": 2.9647361647361645, + "grad_norm": 0.03119639941554429, + "learning_rate": 5.971413262051392e-05, + "loss": 0.0007, + "step": 5760 + }, + { + "epoch": 2.96988416988417, + "grad_norm": 0.04636152011943533, + "learning_rate": 5.9567228281863316e-05, + "loss": 0.0007, + "step": 5770 + }, + { + "epoch": 2.975032175032175, + "grad_norm": 0.03978029859331779, + "learning_rate": 5.942023816657995e-05, + "loss": 0.0007, + "step": 5780 + }, + { + "epoch": 2.98018018018018, + "grad_norm": 0.03374044679795529, + "learning_rate": 5.9273163592528955e-05, + "loss": 0.0007, + "step": 5790 + }, + { + "epoch": 2.9853281853281852, + "grad_norm": 0.03725303440077902, + "learning_rate": 5.912600587833279e-05, + "loss": 0.0007, + "step": 5800 + }, + { + "epoch": 2.9904761904761905, + "grad_norm": 0.04114767320113239, + "learning_rate": 5.8978766343359285e-05, + "loss": 0.0008, + "step": 5810 + }, + { + "epoch": 2.9956241956241954, + "grad_norm": 0.03202816730974482, + "learning_rate": 5.8831446307709824e-05, + "loss": 0.0007, + "step": 5820 + }, + { + "epoch": 3.0005148005148006, + "grad_norm": 0.02466340925296049, + "learning_rate": 5.8684047092207584e-05, + "loss": 0.0007, + "step": 5830 + }, + { + "epoch": 3.005662805662806, + "grad_norm": 0.030811754600515, + "learning_rate": 5.85365700183856e-05, + "loss": 0.0005, + "step": 5840 + }, + { + "epoch": 3.0108108108108107, + "grad_norm": 0.033023021306985235, + "learning_rate": 5.838901640847498e-05, + "loss": 0.0005, + "step": 5850 + }, + { + "epoch": 3.015958815958816, + "grad_norm": 0.0340673351114212, + "learning_rate": 5.8241387585393034e-05, + "loss": 0.0006, + "step": 5860 + }, + { + "epoch": 3.0211068211068213, + "grad_norm": 0.032272959338569815, + "learning_rate": 5.809368487273137e-05, + "loss": 0.0005, + "step": 5870 + }, + { + "epoch": 3.026254826254826, + "grad_norm": 0.02939362185555329, + "learning_rate": 5.794590959474414e-05, + "loss": 0.0005, + "step": 5880 + }, + { + "epoch": 3.0314028314028314, + "grad_norm": 0.04393080385000357, + "learning_rate": 5.779806307633601e-05, + "loss": 0.0005, + "step": 5890 + }, + { + "epoch": 3.0365508365508367, + "grad_norm": 0.03088342648942861, + "learning_rate": 5.765014664305042e-05, + "loss": 0.0005, + "step": 5900 + }, + { + "epoch": 3.0416988416988415, + "grad_norm": 0.0339106097866909, + "learning_rate": 5.7502161621057615e-05, + "loss": 0.0005, + "step": 5910 + }, + { + "epoch": 3.046846846846847, + "grad_norm": 0.03640918824368944, + "learning_rate": 5.73541093371428e-05, + "loss": 0.0004, + "step": 5920 + }, + { + "epoch": 3.051994851994852, + "grad_norm": 0.03729064437085259, + "learning_rate": 5.720599111869425e-05, + "loss": 0.0005, + "step": 5930 + }, + { + "epoch": 3.057142857142857, + "grad_norm": 0.022895579214924446, + "learning_rate": 5.70578082936913e-05, + "loss": 0.0005, + "step": 5940 + }, + { + "epoch": 3.0622908622908622, + "grad_norm": 0.022319582080733442, + "learning_rate": 5.690956219069265e-05, + "loss": 0.0005, + "step": 5950 + }, + { + "epoch": 3.0674388674388675, + "grad_norm": 0.023072663707038983, + "learning_rate": 5.676125413882424e-05, + "loss": 0.0005, + "step": 5960 + }, + { + "epoch": 3.072586872586873, + "grad_norm": 0.04170094514447023, + "learning_rate": 5.661288546776744e-05, + "loss": 0.0005, + "step": 5970 + }, + { + "epoch": 3.0777348777348776, + "grad_norm": 0.037732218375615384, + "learning_rate": 5.6464457507747134e-05, + "loss": 0.0005, + "step": 5980 + }, + { + "epoch": 3.082882882882883, + "grad_norm": 0.034270401550735054, + "learning_rate": 5.631597158951977e-05, + "loss": 0.0005, + "step": 5990 + }, + { + "epoch": 3.088030888030888, + "grad_norm": 0.032494662597365276, + "learning_rate": 5.61674290443614e-05, + "loss": 0.0005, + "step": 6000 + }, + { + "epoch": 3.093178893178893, + "grad_norm": 0.039455279970551045, + "learning_rate": 5.601883120405579e-05, + "loss": 0.0005, + "step": 6010 + }, + { + "epoch": 3.0983268983268983, + "grad_norm": 0.031910010653379776, + "learning_rate": 5.5870179400882495e-05, + "loss": 0.0005, + "step": 6020 + }, + { + "epoch": 3.1034749034749036, + "grad_norm": 0.025824779016766636, + "learning_rate": 5.5721474967604835e-05, + "loss": 0.0005, + "step": 6030 + }, + { + "epoch": 3.1086229086229085, + "grad_norm": 0.04615386842051337, + "learning_rate": 5.5572719237458026e-05, + "loss": 0.0005, + "step": 6040 + }, + { + "epoch": 3.1137709137709138, + "grad_norm": 0.04864194039718604, + "learning_rate": 5.542391354413716e-05, + "loss": 0.0005, + "step": 6050 + }, + { + "epoch": 3.118918918918919, + "grad_norm": 0.033876230012177114, + "learning_rate": 5.5275059221785355e-05, + "loss": 0.0005, + "step": 6060 + }, + { + "epoch": 3.124066924066924, + "grad_norm": 0.03609064764384988, + "learning_rate": 5.512615760498163e-05, + "loss": 0.0006, + "step": 6070 + }, + { + "epoch": 3.129214929214929, + "grad_norm": 0.04954673095430211, + "learning_rate": 5.49772100287291e-05, + "loss": 0.0007, + "step": 6080 + }, + { + "epoch": 3.1343629343629344, + "grad_norm": 0.0583263835485315, + "learning_rate": 5.4828217828442896e-05, + "loss": 0.0008, + "step": 6090 + }, + { + "epoch": 3.1395109395109397, + "grad_norm": 0.04836147185804369, + "learning_rate": 5.467918233993825e-05, + "loss": 0.0008, + "step": 6100 + }, + { + "epoch": 3.1446589446589446, + "grad_norm": 0.03264484167446128, + "learning_rate": 5.453010489941852e-05, + "loss": 0.0006, + "step": 6110 + }, + { + "epoch": 3.14980694980695, + "grad_norm": 0.03178952087355976, + "learning_rate": 5.438098684346313e-05, + "loss": 0.0006, + "step": 6120 + }, + { + "epoch": 3.154954954954955, + "grad_norm": 0.030408107296080268, + "learning_rate": 5.423182950901575e-05, + "loss": 0.0005, + "step": 6130 + }, + { + "epoch": 3.16010296010296, + "grad_norm": 0.041418169894607046, + "learning_rate": 5.40826342333721e-05, + "loss": 0.0006, + "step": 6140 + }, + { + "epoch": 3.1652509652509653, + "grad_norm": 0.043804269456443286, + "learning_rate": 5.393340235416813e-05, + "loss": 0.0006, + "step": 6150 + }, + { + "epoch": 3.1703989703989706, + "grad_norm": 0.03221749562640859, + "learning_rate": 5.378413520936796e-05, + "loss": 0.0005, + "step": 6160 + }, + { + "epoch": 3.1755469755469754, + "grad_norm": 0.030868629433583428, + "learning_rate": 5.363483413725189e-05, + "loss": 0.0005, + "step": 6170 + }, + { + "epoch": 3.1806949806949807, + "grad_norm": 0.026210655424924182, + "learning_rate": 5.348550047640437e-05, + "loss": 0.0005, + "step": 6180 + }, + { + "epoch": 3.185842985842986, + "grad_norm": 0.026229008967355694, + "learning_rate": 5.333613556570208e-05, + "loss": 0.0006, + "step": 6190 + }, + { + "epoch": 3.190990990990991, + "grad_norm": 0.03376442041917258, + "learning_rate": 5.318674074430183e-05, + "loss": 0.0005, + "step": 6200 + }, + { + "epoch": 3.196138996138996, + "grad_norm": 0.02739941790281425, + "learning_rate": 5.3037317351628646e-05, + "loss": 0.0005, + "step": 6210 + }, + { + "epoch": 3.2012870012870014, + "grad_norm": 0.02274292704031371, + "learning_rate": 5.2887866727363664e-05, + "loss": 0.0005, + "step": 6220 + }, + { + "epoch": 3.2064350064350062, + "grad_norm": 0.026244833068079884, + "learning_rate": 5.273839021143218e-05, + "loss": 0.0005, + "step": 6230 + }, + { + "epoch": 3.2115830115830115, + "grad_norm": 0.03313711440917062, + "learning_rate": 5.258888914399168e-05, + "loss": 0.0005, + "step": 6240 + }, + { + "epoch": 3.216731016731017, + "grad_norm": 0.038326250036341686, + "learning_rate": 5.2439364865419715e-05, + "loss": 0.0005, + "step": 6250 + }, + { + "epoch": 3.221879021879022, + "grad_norm": 0.03572938473085202, + "learning_rate": 5.228981871630194e-05, + "loss": 0.0005, + "step": 6260 + }, + { + "epoch": 3.227027027027027, + "grad_norm": 0.04475667953748216, + "learning_rate": 5.214025203742012e-05, + "loss": 0.0004, + "step": 6270 + }, + { + "epoch": 3.232175032175032, + "grad_norm": 0.03180608718141442, + "learning_rate": 5.1990666169740074e-05, + "loss": 0.0005, + "step": 6280 + }, + { + "epoch": 3.2373230373230375, + "grad_norm": 0.04378875425665929, + "learning_rate": 5.184106245439967e-05, + "loss": 0.0005, + "step": 6290 + }, + { + "epoch": 3.2424710424710423, + "grad_norm": 0.02924482520376771, + "learning_rate": 5.169144223269674e-05, + "loss": 0.0004, + "step": 6300 + }, + { + "epoch": 3.2476190476190476, + "grad_norm": 0.03697782009303346, + "learning_rate": 5.1541806846077214e-05, + "loss": 0.0004, + "step": 6310 + }, + { + "epoch": 3.252767052767053, + "grad_norm": 0.031299087522612734, + "learning_rate": 5.139215763612286e-05, + "loss": 0.0005, + "step": 6320 + }, + { + "epoch": 3.2579150579150578, + "grad_norm": 0.025519561592626624, + "learning_rate": 5.124249594453948e-05, + "loss": 0.0004, + "step": 6330 + }, + { + "epoch": 3.263063063063063, + "grad_norm": 0.022158576969015853, + "learning_rate": 5.1092823113144726e-05, + "loss": 0.0004, + "step": 6340 + }, + { + "epoch": 3.2682110682110683, + "grad_norm": 0.03405692010127983, + "learning_rate": 5.094314048385615e-05, + "loss": 0.0004, + "step": 6350 + }, + { + "epoch": 3.273359073359073, + "grad_norm": 0.029263633880294102, + "learning_rate": 5.079344939867916e-05, + "loss": 0.0004, + "step": 6360 + }, + { + "epoch": 3.2785070785070785, + "grad_norm": 0.029131157751809453, + "learning_rate": 5.064375119969491e-05, + "loss": 0.0004, + "step": 6370 + }, + { + "epoch": 3.2836550836550837, + "grad_norm": 0.028045654435947594, + "learning_rate": 5.0494047229048426e-05, + "loss": 0.0005, + "step": 6380 + }, + { + "epoch": 3.288803088803089, + "grad_norm": 0.03125898825372038, + "learning_rate": 5.034433882893644e-05, + "loss": 0.0005, + "step": 6390 + }, + { + "epoch": 3.293951093951094, + "grad_norm": 0.03973619450362519, + "learning_rate": 5.0194627341595366e-05, + "loss": 0.0004, + "step": 6400 + }, + { + "epoch": 3.299099099099099, + "grad_norm": 0.053086571539645445, + "learning_rate": 5.0044914109289345e-05, + "loss": 0.0004, + "step": 6410 + }, + { + "epoch": 3.3042471042471044, + "grad_norm": 0.023336838694777755, + "learning_rate": 4.989520047429814e-05, + "loss": 0.0005, + "step": 6420 + }, + { + "epoch": 3.3093951093951093, + "grad_norm": 0.04926529845051325, + "learning_rate": 4.974548777890516e-05, + "loss": 0.0004, + "step": 6430 + }, + { + "epoch": 3.3145431145431146, + "grad_norm": 0.032291051120605534, + "learning_rate": 4.9595777365385294e-05, + "loss": 0.0005, + "step": 6440 + }, + { + "epoch": 3.31969111969112, + "grad_norm": 0.031231849164216455, + "learning_rate": 4.944607057599309e-05, + "loss": 0.0005, + "step": 6450 + }, + { + "epoch": 3.3248391248391247, + "grad_norm": 0.029205752360175075, + "learning_rate": 4.929636875295052e-05, + "loss": 0.0005, + "step": 6460 + }, + { + "epoch": 3.32998712998713, + "grad_norm": 0.025707129216735455, + "learning_rate": 4.914667323843508e-05, + "loss": 0.0004, + "step": 6470 + }, + { + "epoch": 3.3351351351351353, + "grad_norm": 0.029997846100113502, + "learning_rate": 4.899698537456769e-05, + "loss": 0.0004, + "step": 6480 + }, + { + "epoch": 3.34028314028314, + "grad_norm": 0.020272147316666985, + "learning_rate": 4.8847306503400644e-05, + "loss": 0.0004, + "step": 6490 + }, + { + "epoch": 3.3454311454311454, + "grad_norm": 0.02390116953553025, + "learning_rate": 4.8697637966905654e-05, + "loss": 0.0004, + "step": 6500 + }, + { + "epoch": 3.3505791505791507, + "grad_norm": 0.02649186417157072, + "learning_rate": 4.854798110696177e-05, + "loss": 0.0004, + "step": 6510 + }, + { + "epoch": 3.3557271557271555, + "grad_norm": 0.039386834779333656, + "learning_rate": 4.839833726534332e-05, + "loss": 0.0005, + "step": 6520 + }, + { + "epoch": 3.360875160875161, + "grad_norm": 0.030292353682198625, + "learning_rate": 4.824870778370796e-05, + "loss": 0.0006, + "step": 6530 + }, + { + "epoch": 3.366023166023166, + "grad_norm": 0.0372882456828683, + "learning_rate": 4.809909400358457e-05, + "loss": 0.0004, + "step": 6540 + }, + { + "epoch": 3.371171171171171, + "grad_norm": 0.03478832005171598, + "learning_rate": 4.794949726636123e-05, + "loss": 0.0005, + "step": 6550 + }, + { + "epoch": 3.376319176319176, + "grad_norm": 0.019682849789212142, + "learning_rate": 4.77999189132733e-05, + "loss": 0.0004, + "step": 6560 + }, + { + "epoch": 3.3814671814671815, + "grad_norm": 0.033963568380695516, + "learning_rate": 4.7650360285391235e-05, + "loss": 0.0004, + "step": 6570 + }, + { + "epoch": 3.386615186615187, + "grad_norm": 0.0355596063531261, + "learning_rate": 4.75008227236087e-05, + "loss": 0.0005, + "step": 6580 + }, + { + "epoch": 3.3917631917631916, + "grad_norm": 0.02521070772931787, + "learning_rate": 4.7351307568630417e-05, + "loss": 0.0004, + "step": 6590 + }, + { + "epoch": 3.396911196911197, + "grad_norm": 0.03057740556587922, + "learning_rate": 4.720181616096027e-05, + "loss": 0.0004, + "step": 6600 + }, + { + "epoch": 3.402059202059202, + "grad_norm": 0.017539914127907123, + "learning_rate": 4.705234984088923e-05, + "loss": 0.0004, + "step": 6610 + }, + { + "epoch": 3.407207207207207, + "grad_norm": 0.026289947393918375, + "learning_rate": 4.690290994848332e-05, + "loss": 0.0004, + "step": 6620 + }, + { + "epoch": 3.4123552123552123, + "grad_norm": 0.03174012804775738, + "learning_rate": 4.675349782357166e-05, + "loss": 0.0004, + "step": 6630 + }, + { + "epoch": 3.4175032175032176, + "grad_norm": 0.03160516629332281, + "learning_rate": 4.6604114805734346e-05, + "loss": 0.0004, + "step": 6640 + }, + { + "epoch": 3.4226512226512225, + "grad_norm": 0.019662879337889735, + "learning_rate": 4.6454762234290564e-05, + "loss": 0.0003, + "step": 6650 + }, + { + "epoch": 3.4277992277992277, + "grad_norm": 0.019330726803758146, + "learning_rate": 4.630544144828652e-05, + "loss": 0.0004, + "step": 6660 + }, + { + "epoch": 3.432947232947233, + "grad_norm": 0.03564795667658725, + "learning_rate": 4.615615378648342e-05, + "loss": 0.0005, + "step": 6670 + }, + { + "epoch": 3.4380952380952383, + "grad_norm": 0.030409003871728436, + "learning_rate": 4.6006900587345556e-05, + "loss": 0.0005, + "step": 6680 + }, + { + "epoch": 3.443243243243243, + "grad_norm": 0.02874646517632438, + "learning_rate": 4.5857683189028135e-05, + "loss": 0.0005, + "step": 6690 + }, + { + "epoch": 3.4483912483912484, + "grad_norm": 0.03876243989722429, + "learning_rate": 4.570850292936544e-05, + "loss": 0.0004, + "step": 6700 + }, + { + "epoch": 3.4535392535392537, + "grad_norm": 0.037099282542568675, + "learning_rate": 4.555936114585881e-05, + "loss": 0.0005, + "step": 6710 + }, + { + "epoch": 3.4586872586872586, + "grad_norm": 0.03551751055429998, + "learning_rate": 4.541025917566458e-05, + "loss": 0.0004, + "step": 6720 + }, + { + "epoch": 3.463835263835264, + "grad_norm": 0.02185402984381837, + "learning_rate": 4.5261198355582135e-05, + "loss": 0.0004, + "step": 6730 + }, + { + "epoch": 3.468983268983269, + "grad_norm": 0.057995975629053086, + "learning_rate": 4.511218002204194e-05, + "loss": 0.0005, + "step": 6740 + }, + { + "epoch": 3.474131274131274, + "grad_norm": 0.028945427002689362, + "learning_rate": 4.496320551109352e-05, + "loss": 0.0004, + "step": 6750 + }, + { + "epoch": 3.4792792792792793, + "grad_norm": 0.03749127619966192, + "learning_rate": 4.48142761583935e-05, + "loss": 0.0005, + "step": 6760 + }, + { + "epoch": 3.4844272844272846, + "grad_norm": 0.0280846861564267, + "learning_rate": 4.466539329919365e-05, + "loss": 0.0004, + "step": 6770 + }, + { + "epoch": 3.4895752895752894, + "grad_norm": 0.030846742376742578, + "learning_rate": 4.451655826832887e-05, + "loss": 0.0005, + "step": 6780 + }, + { + "epoch": 3.4947232947232947, + "grad_norm": 0.04288262523046774, + "learning_rate": 4.4367772400205296e-05, + "loss": 0.0005, + "step": 6790 + }, + { + "epoch": 3.4998712998713, + "grad_norm": 0.043988100402121096, + "learning_rate": 4.4219037028788216e-05, + "loss": 0.0005, + "step": 6800 + }, + { + "epoch": 3.505019305019305, + "grad_norm": 0.03211031443753283, + "learning_rate": 4.407035348759022e-05, + "loss": 0.0005, + "step": 6810 + }, + { + "epoch": 3.51016731016731, + "grad_norm": 0.035886182772908656, + "learning_rate": 4.392172310965922e-05, + "loss": 0.0004, + "step": 6820 + }, + { + "epoch": 3.5153153153153154, + "grad_norm": 0.03646471353243471, + "learning_rate": 4.377314722756647e-05, + "loss": 0.0005, + "step": 6830 + }, + { + "epoch": 3.5204633204633202, + "grad_norm": 0.03373800219859811, + "learning_rate": 4.362462717339461e-05, + "loss": 0.0004, + "step": 6840 + }, + { + "epoch": 3.5256113256113255, + "grad_norm": 0.029804816388925213, + "learning_rate": 4.347616427872577e-05, + "loss": 0.0004, + "step": 6850 + }, + { + "epoch": 3.530759330759331, + "grad_norm": 0.02467103605104961, + "learning_rate": 4.3327759874629596e-05, + "loss": 0.0004, + "step": 6860 + }, + { + "epoch": 3.5359073359073356, + "grad_norm": 0.027821577856642653, + "learning_rate": 4.3179415291651336e-05, + "loss": 0.0004, + "step": 6870 + }, + { + "epoch": 3.541055341055341, + "grad_norm": 0.029028021954578164, + "learning_rate": 4.3031131859799896e-05, + "loss": 0.0004, + "step": 6880 + }, + { + "epoch": 3.546203346203346, + "grad_norm": 0.035475237364425344, + "learning_rate": 4.288291090853593e-05, + "loss": 0.0003, + "step": 6890 + }, + { + "epoch": 3.5513513513513515, + "grad_norm": 0.024165396312970457, + "learning_rate": 4.273475376675987e-05, + "loss": 0.0003, + "step": 6900 + }, + { + "epoch": 3.5564993564993563, + "grad_norm": 0.031170339519471815, + "learning_rate": 4.258666176280012e-05, + "loss": 0.0004, + "step": 6910 + }, + { + "epoch": 3.5616473616473616, + "grad_norm": 0.032213709882060224, + "learning_rate": 4.243863622440102e-05, + "loss": 0.0004, + "step": 6920 + }, + { + "epoch": 3.566795366795367, + "grad_norm": 0.032359400970828595, + "learning_rate": 4.229067847871104e-05, + "loss": 0.0004, + "step": 6930 + }, + { + "epoch": 3.571943371943372, + "grad_norm": 0.02858397536544492, + "learning_rate": 4.214278985227087e-05, + "loss": 0.0004, + "step": 6940 + }, + { + "epoch": 3.577091377091377, + "grad_norm": 0.032003333641094575, + "learning_rate": 4.199497167100137e-05, + "loss": 0.0004, + "step": 6950 + }, + { + "epoch": 3.5822393822393823, + "grad_norm": 0.02446093873652105, + "learning_rate": 4.184722526019197e-05, + "loss": 0.0003, + "step": 6960 + }, + { + "epoch": 3.5873873873873876, + "grad_norm": 0.028517373349469785, + "learning_rate": 4.1699551944488535e-05, + "loss": 0.0004, + "step": 6970 + }, + { + "epoch": 3.5925353925353924, + "grad_norm": 0.02112815525395701, + "learning_rate": 4.1551953047881617e-05, + "loss": 0.0003, + "step": 6980 + }, + { + "epoch": 3.5976833976833977, + "grad_norm": 0.024498868593043344, + "learning_rate": 4.140442989369455e-05, + "loss": 0.0003, + "step": 6990 + }, + { + "epoch": 3.602831402831403, + "grad_norm": 0.02958423169210063, + "learning_rate": 4.1256983804571546e-05, + "loss": 0.0004, + "step": 7000 + }, + { + "epoch": 3.607979407979408, + "grad_norm": 0.0313994447951739, + "learning_rate": 4.110961610246594e-05, + "loss": 0.0003, + "step": 7010 + }, + { + "epoch": 3.613127413127413, + "grad_norm": 0.024004164728069384, + "learning_rate": 4.096232810862824e-05, + "loss": 0.0004, + "step": 7020 + }, + { + "epoch": 3.6182754182754184, + "grad_norm": 0.011601574470896738, + "learning_rate": 4.08151211435943e-05, + "loss": 0.0004, + "step": 7030 + }, + { + "epoch": 3.6234234234234233, + "grad_norm": 0.031618517770344294, + "learning_rate": 4.066799652717357e-05, + "loss": 0.0003, + "step": 7040 + }, + { + "epoch": 3.6285714285714286, + "grad_norm": 0.02134447835350743, + "learning_rate": 4.052095557843707e-05, + "loss": 0.0003, + "step": 7050 + }, + { + "epoch": 3.633719433719434, + "grad_norm": 0.03826339638898673, + "learning_rate": 4.037399961570577e-05, + "loss": 0.0004, + "step": 7060 + }, + { + "epoch": 3.6388674388674387, + "grad_norm": 0.03211873138992173, + "learning_rate": 4.022712995653868e-05, + "loss": 0.0004, + "step": 7070 + }, + { + "epoch": 3.644015444015444, + "grad_norm": 0.02466117316522437, + "learning_rate": 4.008034791772101e-05, + "loss": 0.0004, + "step": 7080 + }, + { + "epoch": 3.6491634491634493, + "grad_norm": 0.028574704107980162, + "learning_rate": 3.993365481525242e-05, + "loss": 0.0004, + "step": 7090 + }, + { + "epoch": 3.654311454311454, + "grad_norm": 0.012571808733381363, + "learning_rate": 3.978705196433516e-05, + "loss": 0.0004, + "step": 7100 + }, + { + "epoch": 3.6594594594594594, + "grad_norm": 0.030816425952079327, + "learning_rate": 3.964054067936235e-05, + "loss": 0.0003, + "step": 7110 + }, + { + "epoch": 3.6646074646074647, + "grad_norm": 0.07008208686144377, + "learning_rate": 3.949412227390615e-05, + "loss": 0.0017, + "step": 7120 + }, + { + "epoch": 3.6697554697554695, + "grad_norm": 0.11976556968437012, + "learning_rate": 3.934779806070598e-05, + "loss": 0.0046, + "step": 7130 + }, + { + "epoch": 3.674903474903475, + "grad_norm": 0.061812890673948816, + "learning_rate": 3.9201569351656796e-05, + "loss": 0.0017, + "step": 7140 + }, + { + "epoch": 3.68005148005148, + "grad_norm": 0.03385373999537316, + "learning_rate": 3.905543745779724e-05, + "loss": 0.0009, + "step": 7150 + }, + { + "epoch": 3.685199485199485, + "grad_norm": 0.03564813492069897, + "learning_rate": 3.890940368929799e-05, + "loss": 0.0007, + "step": 7160 + }, + { + "epoch": 3.69034749034749, + "grad_norm": 0.0248508693340784, + "learning_rate": 3.8763469355449934e-05, + "loss": 0.0006, + "step": 7170 + }, + { + "epoch": 3.6954954954954955, + "grad_norm": 0.04188323215145796, + "learning_rate": 3.861763576465249e-05, + "loss": 0.0005, + "step": 7180 + }, + { + "epoch": 3.700643500643501, + "grad_norm": 0.031440435207712, + "learning_rate": 3.8471904224401834e-05, + "loss": 0.0005, + "step": 7190 + }, + { + "epoch": 3.705791505791506, + "grad_norm": 0.03422805009740865, + "learning_rate": 3.832627604127919e-05, + "loss": 0.0005, + "step": 7200 + }, + { + "epoch": 3.710939510939511, + "grad_norm": 0.030483020635419026, + "learning_rate": 3.818075252093909e-05, + "loss": 0.0004, + "step": 7210 + }, + { + "epoch": 3.716087516087516, + "grad_norm": 0.028445665669117794, + "learning_rate": 3.803533496809772e-05, + "loss": 0.0004, + "step": 7220 + }, + { + "epoch": 3.7212355212355215, + "grad_norm": 0.02248715619309222, + "learning_rate": 3.789002468652121e-05, + "loss": 0.0004, + "step": 7230 + }, + { + "epoch": 3.7263835263835263, + "grad_norm": 0.021831289645178505, + "learning_rate": 3.774482297901391e-05, + "loss": 0.0004, + "step": 7240 + }, + { + "epoch": 3.7315315315315316, + "grad_norm": 0.02699270915134675, + "learning_rate": 3.759973114740674e-05, + "loss": 0.0004, + "step": 7250 + }, + { + "epoch": 3.736679536679537, + "grad_norm": 0.026780749973368517, + "learning_rate": 3.745475049254549e-05, + "loss": 0.0004, + "step": 7260 + }, + { + "epoch": 3.7418275418275417, + "grad_norm": 0.03562865071069546, + "learning_rate": 3.7309882314279185e-05, + "loss": 0.0004, + "step": 7270 + }, + { + "epoch": 3.746975546975547, + "grad_norm": 0.025464283056442907, + "learning_rate": 3.7165127911448436e-05, + "loss": 0.0003, + "step": 7280 + }, + { + "epoch": 3.7521235521235523, + "grad_norm": 0.020907122870917453, + "learning_rate": 3.702048858187378e-05, + "loss": 0.0003, + "step": 7290 + }, + { + "epoch": 3.757271557271557, + "grad_norm": 0.02920852210750891, + "learning_rate": 3.687596562234404e-05, + "loss": 0.0003, + "step": 7300 + }, + { + "epoch": 3.7624195624195624, + "grad_norm": 0.043928229435130105, + "learning_rate": 3.673156032860466e-05, + "loss": 0.0004, + "step": 7310 + }, + { + "epoch": 3.7675675675675677, + "grad_norm": 0.02528516790925058, + "learning_rate": 3.65872739953462e-05, + "loss": 0.0003, + "step": 7320 + }, + { + "epoch": 3.7727155727155726, + "grad_norm": 0.025190692936085037, + "learning_rate": 3.644310791619263e-05, + "loss": 0.0004, + "step": 7330 + }, + { + "epoch": 3.777863577863578, + "grad_norm": 0.02949098565167882, + "learning_rate": 3.629906338368975e-05, + "loss": 0.0003, + "step": 7340 + }, + { + "epoch": 3.783011583011583, + "grad_norm": 0.03076761720188141, + "learning_rate": 3.6155141689293635e-05, + "loss": 0.0003, + "step": 7350 + }, + { + "epoch": 3.788159588159588, + "grad_norm": 0.01823526938749411, + "learning_rate": 3.601134412335901e-05, + "loss": 0.0003, + "step": 7360 + }, + { + "epoch": 3.7933075933075933, + "grad_norm": 0.030054979374145634, + "learning_rate": 3.586767197512772e-05, + "loss": 0.0004, + "step": 7370 + }, + { + "epoch": 3.7984555984555985, + "grad_norm": 0.028394157815851345, + "learning_rate": 3.572412653271715e-05, + "loss": 0.0003, + "step": 7380 + }, + { + "epoch": 3.8036036036036034, + "grad_norm": 0.024941003387480946, + "learning_rate": 3.558070908310868e-05, + "loss": 0.0002, + "step": 7390 + }, + { + "epoch": 3.8087516087516087, + "grad_norm": 0.016119334536392298, + "learning_rate": 3.5437420912136186e-05, + "loss": 0.0003, + "step": 7400 + }, + { + "epoch": 3.813899613899614, + "grad_norm": 0.01956011449073766, + "learning_rate": 3.5294263304474406e-05, + "loss": 0.0003, + "step": 7410 + }, + { + "epoch": 3.819047619047619, + "grad_norm": 0.0301432577488622, + "learning_rate": 3.515123754362752e-05, + "loss": 0.0003, + "step": 7420 + }, + { + "epoch": 3.824195624195624, + "grad_norm": 0.02301689966723182, + "learning_rate": 3.5008344911917645e-05, + "loss": 0.0003, + "step": 7430 + }, + { + "epoch": 3.8293436293436294, + "grad_norm": 0.021684817893008985, + "learning_rate": 3.4865586690473275e-05, + "loss": 0.0003, + "step": 7440 + }, + { + "epoch": 3.834491634491634, + "grad_norm": 0.028893022743953608, + "learning_rate": 3.472296415921783e-05, + "loss": 0.0003, + "step": 7450 + }, + { + "epoch": 3.8396396396396395, + "grad_norm": 0.022670241072822814, + "learning_rate": 3.458047859685816e-05, + "loss": 0.0002, + "step": 7460 + }, + { + "epoch": 3.844787644787645, + "grad_norm": 0.01933472388877426, + "learning_rate": 3.443813128087312e-05, + "loss": 0.0003, + "step": 7470 + }, + { + "epoch": 3.84993564993565, + "grad_norm": 0.018473643936503366, + "learning_rate": 3.429592348750208e-05, + "loss": 0.0003, + "step": 7480 + }, + { + "epoch": 3.8550836550836554, + "grad_norm": 0.02136412137946525, + "learning_rate": 3.4153856491733524e-05, + "loss": 0.0003, + "step": 7490 + }, + { + "epoch": 3.86023166023166, + "grad_norm": 0.02235479496968065, + "learning_rate": 3.401193156729356e-05, + "loss": 0.0003, + "step": 7500 + }, + { + "epoch": 3.8653796653796655, + "grad_norm": 0.028540302622685094, + "learning_rate": 3.387014998663453e-05, + "loss": 0.0003, + "step": 7510 + }, + { + "epoch": 3.8705276705276708, + "grad_norm": 0.020128003438604068, + "learning_rate": 3.37285130209236e-05, + "loss": 0.0003, + "step": 7520 + }, + { + "epoch": 3.8756756756756756, + "grad_norm": 0.025538895914475477, + "learning_rate": 3.35870219400314e-05, + "loss": 0.0003, + "step": 7530 + }, + { + "epoch": 3.880823680823681, + "grad_norm": 0.030289979037890078, + "learning_rate": 3.344567801252058e-05, + "loss": 0.0003, + "step": 7540 + }, + { + "epoch": 3.885971685971686, + "grad_norm": 0.019927355309849473, + "learning_rate": 3.3304482505634485e-05, + "loss": 0.0003, + "step": 7550 + }, + { + "epoch": 3.891119691119691, + "grad_norm": 0.02759338057182083, + "learning_rate": 3.31634366852857e-05, + "loss": 0.0003, + "step": 7560 + }, + { + "epoch": 3.8962676962676963, + "grad_norm": 0.028903633467839134, + "learning_rate": 3.3022541816044846e-05, + "loss": 0.0003, + "step": 7570 + }, + { + "epoch": 3.9014157014157016, + "grad_norm": 0.030350690291507897, + "learning_rate": 3.288179916112914e-05, + "loss": 0.0003, + "step": 7580 + }, + { + "epoch": 3.9065637065637064, + "grad_norm": 0.03357795721258163, + "learning_rate": 3.2741209982391084e-05, + "loss": 0.0003, + "step": 7590 + }, + { + "epoch": 3.9117117117117117, + "grad_norm": 0.01695986380953796, + "learning_rate": 3.2600775540307185e-05, + "loss": 0.0003, + "step": 7600 + }, + { + "epoch": 3.916859716859717, + "grad_norm": 0.020896197661077094, + "learning_rate": 3.246049709396659e-05, + "loss": 0.0003, + "step": 7610 + }, + { + "epoch": 3.922007722007722, + "grad_norm": 0.019239127342449314, + "learning_rate": 3.23203759010599e-05, + "loss": 0.0003, + "step": 7620 + }, + { + "epoch": 3.927155727155727, + "grad_norm": 0.024602657539664577, + "learning_rate": 3.218041321786778e-05, + "loss": 0.0003, + "step": 7630 + }, + { + "epoch": 3.9323037323037324, + "grad_norm": 0.025834254295654384, + "learning_rate": 3.2040610299249774e-05, + "loss": 0.0002, + "step": 7640 + }, + { + "epoch": 3.9374517374517373, + "grad_norm": 0.014133882629553975, + "learning_rate": 3.1900968398633035e-05, + "loss": 0.0002, + "step": 7650 + }, + { + "epoch": 3.9425997425997426, + "grad_norm": 0.02270521044377969, + "learning_rate": 3.1761488768001085e-05, + "loss": 0.0002, + "step": 7660 + }, + { + "epoch": 3.947747747747748, + "grad_norm": 0.013036170258669088, + "learning_rate": 3.162217265788252e-05, + "loss": 0.0002, + "step": 7670 + }, + { + "epoch": 3.9528957528957527, + "grad_norm": 0.027603220254549907, + "learning_rate": 3.1483021317339966e-05, + "loss": 0.0003, + "step": 7680 + }, + { + "epoch": 3.958043758043758, + "grad_norm": 0.02664718036129576, + "learning_rate": 3.134403599395871e-05, + "loss": 0.0003, + "step": 7690 + }, + { + "epoch": 3.9631917631917633, + "grad_norm": 0.019228352176302976, + "learning_rate": 3.120521793383562e-05, + "loss": 0.0003, + "step": 7700 + }, + { + "epoch": 3.968339768339768, + "grad_norm": 0.01578358389889038, + "learning_rate": 3.1066568381567917e-05, + "loss": 0.0002, + "step": 7710 + }, + { + "epoch": 3.9734877734877734, + "grad_norm": 0.019143987084199187, + "learning_rate": 3.092808858024202e-05, + "loss": 0.0002, + "step": 7720 + }, + { + "epoch": 3.9786357786357787, + "grad_norm": 0.014749216984668629, + "learning_rate": 3.0789779771422453e-05, + "loss": 0.0002, + "step": 7730 + }, + { + "epoch": 3.983783783783784, + "grad_norm": 0.014687857718271859, + "learning_rate": 3.065164319514066e-05, + "loss": 0.0002, + "step": 7740 + }, + { + "epoch": 3.988931788931789, + "grad_norm": 0.01881418134431877, + "learning_rate": 3.0513680089883885e-05, + "loss": 0.0002, + "step": 7750 + }, + { + "epoch": 3.994079794079794, + "grad_norm": 0.016139007988275365, + "learning_rate": 3.0375891692584157e-05, + "loss": 0.0002, + "step": 7760 + }, + { + "epoch": 3.9992277992277994, + "grad_norm": 0.01677640404443844, + "learning_rate": 3.023827923860703e-05, + "loss": 0.0002, + "step": 7770 + }, + { + "epoch": 4.004118404118404, + "grad_norm": 0.025811797319734466, + "learning_rate": 3.0100843961740672e-05, + "loss": 0.0002, + "step": 7780 + }, + { + "epoch": 4.009266409266409, + "grad_norm": 0.012808158764181392, + "learning_rate": 2.996358709418474e-05, + "loss": 0.0002, + "step": 7790 + }, + { + "epoch": 4.014414414414414, + "grad_norm": 0.006717967648499643, + "learning_rate": 2.982650986653931e-05, + "loss": 0.0002, + "step": 7800 + }, + { + "epoch": 4.01956241956242, + "grad_norm": 0.024326804864823887, + "learning_rate": 2.968961350779389e-05, + "loss": 0.0002, + "step": 7810 + }, + { + "epoch": 4.024710424710425, + "grad_norm": 0.019148557333882772, + "learning_rate": 2.9552899245316324e-05, + "loss": 0.0001, + "step": 7820 + }, + { + "epoch": 4.02985842985843, + "grad_norm": 0.002136352036852221, + "learning_rate": 2.94163683048419e-05, + "loss": 0.0001, + "step": 7830 + }, + { + "epoch": 4.035006435006435, + "grad_norm": 0.017914652753795928, + "learning_rate": 2.9280021910462286e-05, + "loss": 0.0001, + "step": 7840 + }, + { + "epoch": 4.04015444015444, + "grad_norm": 0.006029190293903078, + "learning_rate": 2.9143861284614536e-05, + "loss": 0.0001, + "step": 7850 + }, + { + "epoch": 4.045302445302445, + "grad_norm": 0.01689523464287082, + "learning_rate": 2.900788764807022e-05, + "loss": 0.0001, + "step": 7860 + }, + { + "epoch": 4.050450450450451, + "grad_norm": 0.01668238338763278, + "learning_rate": 2.8872102219924363e-05, + "loss": 0.0001, + "step": 7870 + }, + { + "epoch": 4.055598455598456, + "grad_norm": 0.026463324565300244, + "learning_rate": 2.8736506217584636e-05, + "loss": 0.0001, + "step": 7880 + }, + { + "epoch": 4.0607464607464605, + "grad_norm": 0.013792447015467243, + "learning_rate": 2.860110085676031e-05, + "loss": 0.0001, + "step": 7890 + }, + { + "epoch": 4.065894465894466, + "grad_norm": 0.025858041136368293, + "learning_rate": 2.846588735145152e-05, + "loss": 0.0001, + "step": 7900 + }, + { + "epoch": 4.071042471042471, + "grad_norm": 0.01486015336821821, + "learning_rate": 2.8330866913938196e-05, + "loss": 0.0001, + "step": 7910 + }, + { + "epoch": 4.076190476190476, + "grad_norm": 0.01790013401868239, + "learning_rate": 2.8196040754769316e-05, + "loss": 0.0001, + "step": 7920 + }, + { + "epoch": 4.081338481338482, + "grad_norm": 0.023012184826240983, + "learning_rate": 2.8061410082752072e-05, + "loss": 0.0002, + "step": 7930 + }, + { + "epoch": 4.0864864864864865, + "grad_norm": 0.018039343730181456, + "learning_rate": 2.7926976104940884e-05, + "loss": 0.0001, + "step": 7940 + }, + { + "epoch": 4.091634491634491, + "grad_norm": 0.022021859940559087, + "learning_rate": 2.7792740026626757e-05, + "loss": 0.0002, + "step": 7950 + }, + { + "epoch": 4.096782496782497, + "grad_norm": 0.021636722102628934, + "learning_rate": 2.7658703051326417e-05, + "loss": 0.0002, + "step": 7960 + }, + { + "epoch": 4.101930501930502, + "grad_norm": 0.02173425413911579, + "learning_rate": 2.7524866380771364e-05, + "loss": 0.0001, + "step": 7970 + }, + { + "epoch": 4.107078507078507, + "grad_norm": 0.008324877239857177, + "learning_rate": 2.7391231214897362e-05, + "loss": 0.0001, + "step": 7980 + }, + { + "epoch": 4.1122265122265125, + "grad_norm": 0.02122813596564507, + "learning_rate": 2.7257798751833515e-05, + "loss": 0.0001, + "step": 7990 + }, + { + "epoch": 4.117374517374517, + "grad_norm": 0.0220171287427654, + "learning_rate": 2.7124570187891517e-05, + "loss": 0.0001, + "step": 8000 + }, + { + "epoch": 4.122522522522522, + "grad_norm": 0.02280419073775473, + "learning_rate": 2.699154671755504e-05, + "loss": 0.0001, + "step": 8010 + }, + { + "epoch": 4.127670527670528, + "grad_norm": 0.022268305471070608, + "learning_rate": 2.685872953346889e-05, + "loss": 0.0001, + "step": 8020 + }, + { + "epoch": 4.132818532818533, + "grad_norm": 0.024031026967381568, + "learning_rate": 2.6726119826428397e-05, + "loss": 0.0001, + "step": 8030 + }, + { + "epoch": 4.137966537966538, + "grad_norm": 0.013152365283561488, + "learning_rate": 2.659371878536876e-05, + "loss": 0.0001, + "step": 8040 + }, + { + "epoch": 4.143114543114543, + "grad_norm": 0.016855592394830695, + "learning_rate": 2.6461527597354273e-05, + "loss": 0.0001, + "step": 8050 + }, + { + "epoch": 4.148262548262548, + "grad_norm": 0.005672754988633065, + "learning_rate": 2.6329547447567836e-05, + "loss": 0.0001, + "step": 8060 + }, + { + "epoch": 4.153410553410553, + "grad_norm": 0.010541086264196798, + "learning_rate": 2.6197779519300204e-05, + "loss": 0.0001, + "step": 8070 + }, + { + "epoch": 4.158558558558559, + "grad_norm": 0.019319916187912203, + "learning_rate": 2.606622499393939e-05, + "loss": 0.0002, + "step": 8080 + }, + { + "epoch": 4.163706563706564, + "grad_norm": 0.021368233461297337, + "learning_rate": 2.5934885050960182e-05, + "loss": 0.0001, + "step": 8090 + }, + { + "epoch": 4.168854568854568, + "grad_norm": 0.0325702078610137, + "learning_rate": 2.5803760867913474e-05, + "loss": 0.0002, + "step": 8100 + }, + { + "epoch": 4.174002574002574, + "grad_norm": 0.019137351583682183, + "learning_rate": 2.567285362041567e-05, + "loss": 0.0001, + "step": 8110 + }, + { + "epoch": 4.179150579150579, + "grad_norm": 0.008408227112363241, + "learning_rate": 2.5542164482138288e-05, + "loss": 0.0001, + "step": 8120 + }, + { + "epoch": 4.184298584298585, + "grad_norm": 0.011824229857915829, + "learning_rate": 2.5411694624797265e-05, + "loss": 0.0002, + "step": 8130 + }, + { + "epoch": 4.1894465894465895, + "grad_norm": 0.012378838152749616, + "learning_rate": 2.5281445218142573e-05, + "loss": 0.0001, + "step": 8140 + }, + { + "epoch": 4.194594594594594, + "grad_norm": 0.008348734044372443, + "learning_rate": 2.5151417429947745e-05, + "loss": 0.0001, + "step": 8150 + }, + { + "epoch": 4.1997425997426, + "grad_norm": 0.010210360750415117, + "learning_rate": 2.5021612425999246e-05, + "loss": 0.0001, + "step": 8160 + }, + { + "epoch": 4.204890604890605, + "grad_norm": 0.01661505328659959, + "learning_rate": 2.4892031370086248e-05, + "loss": 0.0001, + "step": 8170 + }, + { + "epoch": 4.21003861003861, + "grad_norm": 0.010547341849269087, + "learning_rate": 2.4762675423990007e-05, + "loss": 0.0001, + "step": 8180 + }, + { + "epoch": 4.2151866151866155, + "grad_norm": 0.01811505314576, + "learning_rate": 2.4633545747473508e-05, + "loss": 0.0002, + "step": 8190 + }, + { + "epoch": 4.22033462033462, + "grad_norm": 0.010236309343466773, + "learning_rate": 2.4504643498271123e-05, + "loss": 0.0001, + "step": 8200 + }, + { + "epoch": 4.225482625482625, + "grad_norm": 0.02260238253027528, + "learning_rate": 2.4375969832078188e-05, + "loss": 0.0001, + "step": 8210 + }, + { + "epoch": 4.230630630630631, + "grad_norm": 0.01898656708411786, + "learning_rate": 2.42475259025406e-05, + "loss": 0.0001, + "step": 8220 + }, + { + "epoch": 4.235778635778636, + "grad_norm": 0.0216466816156467, + "learning_rate": 2.4119312861244507e-05, + "loss": 0.0001, + "step": 8230 + }, + { + "epoch": 4.240926640926641, + "grad_norm": 0.013534955504234794, + "learning_rate": 2.399133185770605e-05, + "loss": 0.0001, + "step": 8240 + }, + { + "epoch": 4.246074646074646, + "grad_norm": 0.018107142241368435, + "learning_rate": 2.386358403936092e-05, + "loss": 0.0001, + "step": 8250 + }, + { + "epoch": 4.251222651222651, + "grad_norm": 0.020118284339340806, + "learning_rate": 2.3736070551554217e-05, + "loss": 0.0001, + "step": 8260 + }, + { + "epoch": 4.256370656370656, + "grad_norm": 0.012741350477805358, + "learning_rate": 2.360879253753006e-05, + "loss": 0.0001, + "step": 8270 + }, + { + "epoch": 4.261518661518662, + "grad_norm": 0.010132247778619484, + "learning_rate": 2.3481751138421388e-05, + "loss": 0.0001, + "step": 8280 + }, + { + "epoch": 4.266666666666667, + "grad_norm": 0.01271003449347417, + "learning_rate": 2.335494749323977e-05, + "loss": 0.0001, + "step": 8290 + }, + { + "epoch": 4.2718146718146714, + "grad_norm": 0.031203860406014736, + "learning_rate": 2.3228382738865102e-05, + "loss": 0.0001, + "step": 8300 + }, + { + "epoch": 4.276962676962677, + "grad_norm": 0.010318940080235485, + "learning_rate": 2.310205801003549e-05, + "loss": 0.0001, + "step": 8310 + }, + { + "epoch": 4.282110682110682, + "grad_norm": 0.018595944803491457, + "learning_rate": 2.2975974439337088e-05, + "loss": 0.0001, + "step": 8320 + }, + { + "epoch": 4.287258687258687, + "grad_norm": 0.011506108232832438, + "learning_rate": 2.2850133157193787e-05, + "loss": 0.0001, + "step": 8330 + }, + { + "epoch": 4.292406692406693, + "grad_norm": 0.014557942584659167, + "learning_rate": 2.272453529185732e-05, + "loss": 0.0001, + "step": 8340 + }, + { + "epoch": 4.297554697554697, + "grad_norm": 0.0124496461790406, + "learning_rate": 2.2599181969396987e-05, + "loss": 0.0001, + "step": 8350 + }, + { + "epoch": 4.302702702702703, + "grad_norm": 0.013148251144479798, + "learning_rate": 2.2474074313689553e-05, + "loss": 0.0001, + "step": 8360 + }, + { + "epoch": 4.307850707850708, + "grad_norm": 0.025248251579773607, + "learning_rate": 2.2349213446409296e-05, + "loss": 0.0001, + "step": 8370 + }, + { + "epoch": 4.312998712998713, + "grad_norm": 0.010216130756651946, + "learning_rate": 2.2224600487017792e-05, + "loss": 0.0001, + "step": 8380 + }, + { + "epoch": 4.318146718146718, + "grad_norm": 0.012406597553462226, + "learning_rate": 2.2100236552753983e-05, + "loss": 0.0001, + "step": 8390 + }, + { + "epoch": 4.323294723294723, + "grad_norm": 0.015400236448409597, + "learning_rate": 2.197612275862417e-05, + "loss": 0.0001, + "step": 8400 + }, + { + "epoch": 4.328442728442728, + "grad_norm": 0.017483326328530974, + "learning_rate": 2.1852260217391912e-05, + "loss": 0.0001, + "step": 8410 + }, + { + "epoch": 4.333590733590734, + "grad_norm": 0.010856789846218906, + "learning_rate": 2.1728650039568193e-05, + "loss": 0.0001, + "step": 8420 + }, + { + "epoch": 4.338738738738739, + "grad_norm": 0.008868225000820098, + "learning_rate": 2.160529333340133e-05, + "loss": 0.0001, + "step": 8430 + }, + { + "epoch": 4.343886743886744, + "grad_norm": 0.01802504248778034, + "learning_rate": 2.1482191204867096e-05, + "loss": 0.0001, + "step": 8440 + }, + { + "epoch": 4.349034749034749, + "grad_norm": 0.037614846123158796, + "learning_rate": 2.1359344757658846e-05, + "loss": 0.0001, + "step": 8450 + }, + { + "epoch": 4.354182754182754, + "grad_norm": 0.015718171372672392, + "learning_rate": 2.1236755093177575e-05, + "loss": 0.0001, + "step": 8460 + }, + { + "epoch": 4.359330759330759, + "grad_norm": 0.020407369069486658, + "learning_rate": 2.1114423310522013e-05, + "loss": 0.0001, + "step": 8470 + }, + { + "epoch": 4.364478764478765, + "grad_norm": 0.01950379142868458, + "learning_rate": 2.0992350506478813e-05, + "loss": 0.0001, + "step": 8480 + }, + { + "epoch": 4.36962676962677, + "grad_norm": 0.012781684498188879, + "learning_rate": 2.087053777551275e-05, + "loss": 0.0001, + "step": 8490 + }, + { + "epoch": 4.3747747747747745, + "grad_norm": 0.01755124172596203, + "learning_rate": 2.074898620975681e-05, + "loss": 0.0001, + "step": 8500 + }, + { + "epoch": 4.37992277992278, + "grad_norm": 0.015072657490458602, + "learning_rate": 2.0627696899002523e-05, + "loss": 0.0001, + "step": 8510 + }, + { + "epoch": 4.385070785070785, + "grad_norm": 0.010698412547935427, + "learning_rate": 2.0506670930690074e-05, + "loss": 0.0001, + "step": 8520 + }, + { + "epoch": 4.39021879021879, + "grad_norm": 0.013605901511973946, + "learning_rate": 2.0385909389898612e-05, + "loss": 0.0001, + "step": 8530 + }, + { + "epoch": 4.395366795366796, + "grad_norm": 0.0213518477823511, + "learning_rate": 2.026541335933656e-05, + "loss": 0.0001, + "step": 8540 + }, + { + "epoch": 4.4005148005148005, + "grad_norm": 0.007145665368449368, + "learning_rate": 2.0145183919331795e-05, + "loss": 0.0001, + "step": 8550 + }, + { + "epoch": 4.405662805662805, + "grad_norm": 0.025949019450832108, + "learning_rate": 2.0025222147822082e-05, + "loss": 0.0001, + "step": 8560 + }, + { + "epoch": 4.410810810810811, + "grad_norm": 0.011758840605907199, + "learning_rate": 1.9905529120345368e-05, + "loss": 0.0001, + "step": 8570 + }, + { + "epoch": 4.415958815958816, + "grad_norm": 0.006950300275636968, + "learning_rate": 1.9786105910030085e-05, + "loss": 0.0001, + "step": 8580 + }, + { + "epoch": 4.421106821106821, + "grad_norm": 0.011572227158215929, + "learning_rate": 1.9666953587585568e-05, + "loss": 0.0001, + "step": 8590 + }, + { + "epoch": 4.4262548262548265, + "grad_norm": 0.010038678886482718, + "learning_rate": 1.9548073221292517e-05, + "loss": 0.0001, + "step": 8600 + }, + { + "epoch": 4.431402831402831, + "grad_norm": 0.019210210066316586, + "learning_rate": 1.9429465876993307e-05, + "loss": 0.0001, + "step": 8610 + }, + { + "epoch": 4.436550836550836, + "grad_norm": 0.023724435769078244, + "learning_rate": 1.9311132618082533e-05, + "loss": 0.0001, + "step": 8620 + }, + { + "epoch": 4.441698841698842, + "grad_norm": 0.024117442980680612, + "learning_rate": 1.91930745054974e-05, + "loss": 0.0001, + "step": 8630 + }, + { + "epoch": 4.446846846846847, + "grad_norm": 0.012233228850979216, + "learning_rate": 1.9075292597708218e-05, + "loss": 0.0001, + "step": 8640 + }, + { + "epoch": 4.4519948519948525, + "grad_norm": 0.004823196579236481, + "learning_rate": 1.8957787950709005e-05, + "loss": 0.0001, + "step": 8650 + }, + { + "epoch": 4.457142857142857, + "grad_norm": 0.010943171179007932, + "learning_rate": 1.8840561618007873e-05, + "loss": 0.0001, + "step": 8660 + }, + { + "epoch": 4.462290862290862, + "grad_norm": 0.013066188760788387, + "learning_rate": 1.8723614650617723e-05, + "loss": 0.0001, + "step": 8670 + }, + { + "epoch": 4.467438867438867, + "grad_norm": 0.019413613174569196, + "learning_rate": 1.8606948097046755e-05, + "loss": 0.0001, + "step": 8680 + }, + { + "epoch": 4.472586872586873, + "grad_norm": 0.009833573771483015, + "learning_rate": 1.8490563003288986e-05, + "loss": 0.0001, + "step": 8690 + }, + { + "epoch": 4.4777348777348775, + "grad_norm": 0.013780014365617276, + "learning_rate": 1.837446041281504e-05, + "loss": 0.0001, + "step": 8700 + }, + { + "epoch": 4.482882882882883, + "grad_norm": 0.014975404841140355, + "learning_rate": 1.8258641366562717e-05, + "loss": 0.0001, + "step": 8710 + }, + { + "epoch": 4.488030888030888, + "grad_norm": 0.024022134667883607, + "learning_rate": 1.814310690292758e-05, + "loss": 0.0001, + "step": 8720 + }, + { + "epoch": 4.493178893178893, + "grad_norm": 0.005457919384423813, + "learning_rate": 1.80278580577538e-05, + "loss": 0.0001, + "step": 8730 + }, + { + "epoch": 4.498326898326899, + "grad_norm": 0.014348324514181122, + "learning_rate": 1.7912895864324724e-05, + "loss": 0.0001, + "step": 8740 + }, + { + "epoch": 4.5034749034749035, + "grad_norm": 0.01043124211100229, + "learning_rate": 1.7798221353353685e-05, + "loss": 0.0001, + "step": 8750 + }, + { + "epoch": 4.508622908622908, + "grad_norm": 0.017431358224551677, + "learning_rate": 1.768383555297478e-05, + "loss": 0.0001, + "step": 8760 + }, + { + "epoch": 4.513770913770914, + "grad_norm": 0.031046508015317578, + "learning_rate": 1.7569739488733578e-05, + "loss": 0.0001, + "step": 8770 + }, + { + "epoch": 4.518918918918919, + "grad_norm": 0.012100059264824605, + "learning_rate": 1.7455934183578026e-05, + "loss": 0.0001, + "step": 8780 + }, + { + "epoch": 4.524066924066924, + "grad_norm": 0.025906220128756364, + "learning_rate": 1.734242065784917e-05, + "loss": 0.0001, + "step": 8790 + }, + { + "epoch": 4.5292149292149295, + "grad_norm": 0.026461974955074175, + "learning_rate": 1.7229199929272055e-05, + "loss": 0.0001, + "step": 8800 + }, + { + "epoch": 4.534362934362934, + "grad_norm": 0.020784091573625766, + "learning_rate": 1.711627301294663e-05, + "loss": 0.0001, + "step": 8810 + }, + { + "epoch": 4.539510939510939, + "grad_norm": 0.015724773196687265, + "learning_rate": 1.700364092133865e-05, + "loss": 0.0001, + "step": 8820 + }, + { + "epoch": 4.544658944658945, + "grad_norm": 0.022033956058518236, + "learning_rate": 1.6891304664270484e-05, + "loss": 0.0001, + "step": 8830 + }, + { + "epoch": 4.54980694980695, + "grad_norm": 0.012365405701111597, + "learning_rate": 1.67792652489122e-05, + "loss": 0.0001, + "step": 8840 + }, + { + "epoch": 4.554954954954955, + "grad_norm": 0.022817448282821208, + "learning_rate": 1.666752367977249e-05, + "loss": 0.0001, + "step": 8850 + }, + { + "epoch": 4.56010296010296, + "grad_norm": 0.020920092254860793, + "learning_rate": 1.6556080958689607e-05, + "loss": 0.0001, + "step": 8860 + }, + { + "epoch": 4.565250965250965, + "grad_norm": 0.01074061671376947, + "learning_rate": 1.6444938084822485e-05, + "loss": 0.0001, + "step": 8870 + }, + { + "epoch": 4.57039897039897, + "grad_norm": 0.012533064591247882, + "learning_rate": 1.63340960546417e-05, + "loss": 0.0001, + "step": 8880 + }, + { + "epoch": 4.575546975546976, + "grad_norm": 0.0154407840642222, + "learning_rate": 1.6223555861920532e-05, + "loss": 0.0001, + "step": 8890 + }, + { + "epoch": 4.580694980694981, + "grad_norm": 0.008609029008328863, + "learning_rate": 1.6113318497726155e-05, + "loss": 0.0001, + "step": 8900 + }, + { + "epoch": 4.585842985842985, + "grad_norm": 0.02184882891249787, + "learning_rate": 1.6003384950410604e-05, + "loss": 0.0001, + "step": 8910 + }, + { + "epoch": 4.590990990990991, + "grad_norm": 0.016324549126701594, + "learning_rate": 1.5893756205602052e-05, + "loss": 0.0001, + "step": 8920 + }, + { + "epoch": 4.596138996138996, + "grad_norm": 0.0178856318300857, + "learning_rate": 1.5784433246195918e-05, + "loss": 0.0001, + "step": 8930 + }, + { + "epoch": 4.601287001287002, + "grad_norm": 0.015261908534947037, + "learning_rate": 1.5675417052345957e-05, + "loss": 0.0001, + "step": 8940 + }, + { + "epoch": 4.606435006435007, + "grad_norm": 0.02456967807841194, + "learning_rate": 1.556670860145567e-05, + "loss": 0.0001, + "step": 8950 + }, + { + "epoch": 4.611583011583011, + "grad_norm": 0.00975633247678655, + "learning_rate": 1.5458308868169403e-05, + "loss": 0.0001, + "step": 8960 + }, + { + "epoch": 4.616731016731016, + "grad_norm": 0.019573664293844415, + "learning_rate": 1.5350218824363598e-05, + "loss": 0.0001, + "step": 8970 + }, + { + "epoch": 4.621879021879022, + "grad_norm": 0.012394685867736488, + "learning_rate": 1.52424394391382e-05, + "loss": 0.0001, + "step": 8980 + }, + { + "epoch": 4.627027027027027, + "grad_norm": 0.010317846737522209, + "learning_rate": 1.5134971678807824e-05, + "loss": 0.0001, + "step": 8990 + }, + { + "epoch": 4.632175032175033, + "grad_norm": 0.007701510515865398, + "learning_rate": 1.5027816506893172e-05, + "loss": 0.0001, + "step": 9000 + } + ], + "logging_steps": 10, + "max_steps": 11658, + "num_input_tokens_seen": 0, + "num_train_epochs": 6, + "save_steps": 1800, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 807471670099968.0, + "train_batch_size": 128, + "trial_name": null, + "trial_params": null +}